2023-03-25 19:26:40 +01:00
|
|
|
#include "common.h"
|
2023-08-28 17:59:39 +02:00
|
|
|
#include "build-info.h"
|
|
|
|
#include "llama.h"
|
2023-03-24 16:19:05 +01:00
|
|
|
|
2023-08-28 17:59:39 +02:00
|
|
|
#include <algorithm>
|
2023-03-11 00:04:06 +01:00
|
|
|
#include <cassert>
|
2023-08-28 17:59:39 +02:00
|
|
|
#include <cmath>
|
2023-03-11 00:04:06 +01:00
|
|
|
#include <cstring>
|
2023-08-28 17:59:39 +02:00
|
|
|
#include <ctime>
|
2023-03-10 19:40:58 +01:00
|
|
|
#include <fstream>
|
2023-03-22 06:32:36 +01:00
|
|
|
#include <iterator>
|
2023-08-28 17:59:39 +02:00
|
|
|
#include <iostream>
|
|
|
|
#include <regex>
|
llama : new sampling algorithms (#1126)
* Sample interface, new samplers.
New samplers:
- locally typical sampling
- tail free sampling
- frequency and presence penalty
- mirostat
Ignore EOS fix: -inf should be used.
* mirostat
* Added --logit-bias and --no-penalize-nl, removed std::span
* Use C++11, clarify llama API documentation, rename Mirostat parameters to --mirostat_lr and --mirostat_ent, add temperature sampling for Mirostat, simplify Mirostat sampling API parameters (removed N and *k)
Use C++11, clarify llama API documentation, rename Mirostat parameters to --mirostat_lr and --mirostat_ent, add temperature sampling for Mirostat, simplify Mirostat sampling API parameters (removed N and *k)
* Save and load example adjust
* Tests
* Windows build fix
* Windows test fix
2023-04-29 07:34:41 +02:00
|
|
|
#include <sstream>
|
2023-08-28 17:59:39 +02:00
|
|
|
#include <string>
|
2023-05-15 04:25:42 +02:00
|
|
|
#include <unordered_set>
|
2023-08-28 17:59:39 +02:00
|
|
|
#include <vector>
|
train : mem usage and other improvements (#2439)
* fix track_max_mem in forward_batch_wo_cache_flash_attn_train
* remove unnecessary Adam(W) optimizer tensors.
reduces optimizer memory overhead from 7*modelsize to 2*modelsize.
additionally allows to optimize models with more than 2^31 parameters by replacing int with int64_t.
bumps training checkpoint file version, but old checkpoints can still be read.
new version with less tensors is saved.
* add gradient clipping to AdamW
* Fix reset of unused g->nodes and g->grads to NULL
* implement gradient checkpointing for training
reduces memory overhead from O(n_layer) to O(sqrt(n_layer))
as explained in readme of https://github.com/cybertronai/gradient-checkpointing
* remove unused compute buffer 3
* add and use function ggml_build_backward_expand to avoid stack overflows with large maximum number of nodes
GGML_API void ggml_build_backward_expand(struct ggml_context * ctx, struct ggml_cgraph * gf, struct ggml_cgraph * gb, bool keep);
* change AdamW decay parameter to work like the torch AdamW decay parameter
It is now relative to Adam learning rate `alpha*sched`.
Before that it was relative to `sched` only.
`alpha` being the maximum learning rate and `sched` being a scaling parameter in [0..1]
* change default AdamW weight decay parameter used in training to 0.1 as used in nanoGPT
* change default AdamW weight decay parameter defined in ggml to 0.0, making Adam default instead of AdamW
btw: the default weight decay parameter for torch.optim.AdamW is 0.01
* bug fixes for cross entropy loss
ggml_cross_entropy_loss: sums where not correctly added in workload of each thread
ggml_cross_entropy_loss_back: simplify backward process, reducing numerical issues
guard usage of exp f16 lookup in cross entropy by #define GGML_CROSS_ENTROPY_EXP_FP16
cross entropy loss is only used once during training, but it is quite sensitive to numerical errors introduced by exp-f16-lookup.
so exp-f16-lookup for cross entropy loss is disabled by default, trading better gradients for very slightly worse runtime performance.
* fix test-grad0 for cross_entropy_loss
the second argument to cross_entropy_loss must sum up to 1 for each row
* fix test-grad0 for soft_max
dont use only sum as aggregation, because sum of softmax is always 1 -> finite differences should not work
instead use sum(log(soft_max()*(1-eps)+eps)); use eps to avoid log(0)
* improve finite differences of test-grad0 by using double instead of float
* change cross_entropy_loss to output average over all rows
this helps keeping the loss and gradients in a sane range
* improve gradient checkpointing
sqrt(n_layers) is only the best checkpoint step when mem size of checkpoints and mem size of layers are equal.
since layers require more memory than the single-tensor-checkpoint we use, the optimal values are compute different:
```
given: n, u, v
objective: minimize(a*u+b*v) where a*b=n, a>0, b>0
b=n/a
minimize(a*u+v*n/a)
diff(a*u+v*n/a, a) = u - (v*n/a)/a
diff(a*u+v*n/a, a) == 0
u - (v*n/a)/a == 0
u == v*n/(a*a)
u*a*a = v*n
a*a = v*n/u
a = sqrt(n*v/u)
```
this change results in more checkpoints, requiring less layers to store between checkpoints, overall improving memory usage.
* disable gradient checkpointing debug output
* llama : fix rope usage in train-text-from-scratch after ChatGLM change
* add more training parameters:
--enable-restart N Only for Adam optimizer. Enable restarts of cos-decay
--disable-restart N Only for Adam optimizer. Disable restarts of cos-decay
--opt-past N Number of optimization iterations to track for delta convergence test. Disabled when zero.
--opt-delta N Maximum delta for delta convergence test. Disabled when <= zero.
--opt-max-no-improvement N Maximum number of optimization iterations with no improvement. Disabled when <= zero.
--adam-epsf N AdamW epsilon for convergence test. Disabled when <= zero.
--adam-min-alpha N Adam minimum learning rate alpha, usually 0.1 * alpha
* replace memcpy with reshape operation so that the graph is not cut at the input
this makes it possible to store other values into the input tensor and then simply recompute the graph without rebuilding it
* remove unused function argument from get_example_targets_batch
* measure and print total training time
* add optimization callback to ggml_opt_resume_g
this callback is called before each iteration with custom data and pointer to learning schedule parameter (only used in Adam(W)).
can be used for dynamic learning schedule and setting input data for batches before each iteration
* use optimization callback in training
allows dynamic learning schedule and different batch data for each iteration without relying on low n_iter and high n_examples parameters
reduces runtime by avoiding restart of optimization function and improves training convergence by providing a different batch for each iteration
* add minimum number of tensor dimensions to apply weight decay (default 2)
this allows to not apply weight decay to bias parameters
* rename training parameter cos-decay-alpha to cos-decay-min and clarify that adam-min-alpha also applies to warmup
* fix increase of model.train_samples and model.train_tokens
now that each optimizer iteration gets its own batch we need to multiply by number of opt iterations
* change sampling parameters for prediction after training to defaults of common.h
and clarify what is context for prediction and what are generated tokens
* tighten abs error bounds for cross_entropy_loss in test-grad0
* add conditional compilation of using F16 exp in flash attention
uncomment `// #define GGML_FLASH_ATTN_EXP_FP16` to enable usage of f16 exp in flash attention
* tighten abs error bounds for flash_attn in test-grad0
* tighten abs error bounds for sqrt in test-grad0
* remove out-commented vectorized code of opt_adam
the vectorized code might be bit faster for low number of parameters, but it had a big memory usage overhead
* ggml : update ggml_rms_norm_back with configurable eps
* llama training : fix ggml_rms_norm_back calls to pass configurable eps
* remove trailing whitespace
* add train function using automatic gradient checkpointing backward pass and allocator
* in train function replace add_inplace by regular add
because using add_inplace seems to result in different gradients
* don't use allocate hash_map on context
because the context has no_alloc=True when using memory allocator resulting in NULL data pointers
* correctly clone reshape and permute operations by also cloning tensor->nb values
* fix variable name and add missing type cast
* terminate recursive tensor cloning when reaching tensor without src tensors
* correctly clone view tensors by setting data pointers
without this the checkpointing would only work when being used together with memory allocator
* fix variable names
* swap arguments to commutative ops to be the same as in `forward_batch_wo_cache_flash_attn`
* add input tensors as checkpoints
so that recursive tensor cloning of gradient checkpointing terminates on input tensors
* fix variable name and add missing boolean negation
* make sure some tensors are not reallocated by inserting new temporary nodes depending on them:
output and parameter gradient tensors need to be available at the end of the graph execution
parameter gradient tensors also need to be available before the graph execution because they are set to zero before each optimizer iteration
checkpoint tensors are allocated all together to reduce memory allocator fragmentation
afterwards, in addition to the temporary nodes, we also need to reset the temporary leafs
* fix ASSERT to work with zero layers
* add training options whether to use allocator and/or unified training function
* integrate unified training function which may use memory allocator
the unified training function also supports arguments whether to use flash attention and/or gradient checkpointing
* format name of cloned tensors with " (clone)" suffix
* set names for tensors in unified train function for easier debugging
* allocate graph on context using ggml_new_graph
* remove handwritten training functions
* remove unused training parameters "use_scratch" and "use_unified"
* remove trailing whitespace
* remove unused train params: mem_compute1_gb & mem_compute2_gb
mem_compute_gb is used for compute when automatic memory allocator is not enabled, otherwise it can be very small to only hold the tensor definitions
mem_compute0_gb is used for automatic memory allocator (as long as measurement of max required size is not implemented)
* remove unused forward_batch function
* add debug asserts in ggml_allocr_alloc to some common pitfalls when using this function directly
* only use ggml_allocr_alloc when tensor has NULL data and is no view
* fix test when to create temporary backward graph
temporary backward graph is only necessary when using checkpointing
* fix memory "leak" in optimizers
each iteration a new cplan with new memory for work data was allocated.
now cplan creation only happens at the start of optimization, with each iteration reusing the cplan and its work data.
* reverse order of for loop in ggml_build_backward_expand to save memory when using gradient checkpointing and allocator
with this loop order gradient checkpointing with allocator on 16 layer model saves 13% memory; 2 layer memory it saves 2% memory.
the computation results are the same
* add missing lctx argument to get_example_targets_batch
* implement llama model file saving using gguf
checkpoint loading and saving disabled, to be replaced by loading and saving via gguf
* implement loading/saving of checkpointing files using GGUF
* bug fixes
* add checkpoint file version for future compatibility
* update readme with gguf filenames
* save & load opt->just_initialized value
* add first draft for checkpoint conversion script
* add gguf arch and ftype
* save opt parameter counter as uint64
* add gguf key and tensor names for optimizer and training
* add layer_norm_rms_eps to checkpoint convert script
* use same GGUF_GET_KEY macro as in llama.cpp
* use norm_rms_eps, and rope parameters and command line options to set them
* fix memory corruption bug in gguf
ctx->kv and ctx->infos was reallocated using not-aligned realloc, but freed with aligned free.
to fix this a GGML_ALIGNED_REALLOC was added, but there is no posix_memalign_realloc function.
so on non-windows and non-mingw32 platforms we fall back to aligned malloc, followed by copying
and freeing the old data.
* add gguf example cmake file
* bug fixes in tokenize_file
* bug fixes in load_llama_model_gguf
* bug fix: init model when no checkpoint was loaded
* bug fix in read_tensor_by_name
* bug fix in load_opt_context_gguf
* avoid printing lots of spaced on the unusual case that loss gets nan
* set name of tensors with empty name from what was read from gguf
* remove trailing whitespace
* print data checksums before saving and after loading to verify correctness
* bug fixes for convert-train-checkpoint-to-gguf
* temporarily add code to write old checkpoint files
used to verify that old checkpoint files are correctly converted to gguf
* bug fixes for convert-train-checkpoint-to-gguf.py loading checkpoints with opt_version=0
* remove code used to verify correctness of checkpoint file conversion
* remove trailing whitespace
* remove prediction related code
use main for prediction, it is better optimized
* update train-text-from-scratch README.md
* fix non-windows GGML_ALIGNED_REALLOC
* add missing blank line at end of file
* remove GGML_ALIGNED_REALLOC and use normal malloc/realloc/free for gguf ctx->kv & ctx->infos
* train : fix compile warnings
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-08-28 21:51:47 +02:00
|
|
|
#include <cinttypes>
|
2023-04-30 20:41:35 +02:00
|
|
|
|
|
|
|
#if defined(__APPLE__) && defined(__MACH__)
|
|
|
|
#include <sys/types.h>
|
|
|
|
#include <sys/sysctl.h>
|
|
|
|
#endif
|
2023-03-10 19:40:58 +01:00
|
|
|
|
2023-05-09 04:45:48 +02:00
|
|
|
#if defined(_WIN32)
|
|
|
|
#define WIN32_LEAN_AND_MEAN
|
|
|
|
#define NOMINMAX
|
2023-08-28 17:59:39 +02:00
|
|
|
#include <codecvt>
|
|
|
|
#include <locale>
|
2023-05-09 04:45:48 +02:00
|
|
|
#include <windows.h>
|
2023-04-08 17:49:39 +02:00
|
|
|
#include <fcntl.h>
|
|
|
|
#include <io.h>
|
2023-05-09 04:45:48 +02:00
|
|
|
#else
|
|
|
|
#include <sys/ioctl.h>
|
2023-08-28 17:59:39 +02:00
|
|
|
#include <sys/stat.h>
|
2023-05-09 04:45:48 +02:00
|
|
|
#include <unistd.h>
|
2023-03-28 16:09:55 +02:00
|
|
|
#endif
|
2023-03-12 21:15:00 +01:00
|
|
|
|
2023-06-16 20:23:53 +02:00
|
|
|
#if defined(_MSC_VER)
|
|
|
|
#pragma warning(disable: 4244 4267) // possible loss of data
|
|
|
|
#endif
|
|
|
|
|
2023-04-30 20:41:35 +02:00
|
|
|
int32_t get_num_physical_cores() {
|
2023-03-17 18:47:35 +01:00
|
|
|
#ifdef __linux__
|
2023-05-15 04:25:42 +02:00
|
|
|
// enumerate the set of thread siblings, num entries is num cores
|
|
|
|
std::unordered_set<std::string> siblings;
|
|
|
|
for (uint32_t cpu=0; cpu < UINT32_MAX; ++cpu) {
|
|
|
|
std::ifstream thread_siblings("/sys/devices/system/cpu"
|
|
|
|
+ std::to_string(cpu) + "/topology/thread_siblings");
|
|
|
|
if (!thread_siblings.is_open()) {
|
|
|
|
break; // no more cpus
|
2023-04-30 20:41:35 +02:00
|
|
|
}
|
2023-05-15 04:25:42 +02:00
|
|
|
std::string line;
|
|
|
|
if (std::getline(thread_siblings, line)) {
|
|
|
|
siblings.insert(line);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (siblings.size() > 0) {
|
|
|
|
return static_cast<int32_t>(siblings.size());
|
2023-04-30 20:41:35 +02:00
|
|
|
}
|
|
|
|
#elif defined(__APPLE__) && defined(__MACH__)
|
|
|
|
int32_t num_physical_cores;
|
|
|
|
size_t len = sizeof(num_physical_cores);
|
|
|
|
int result = sysctlbyname("hw.perflevel0.physicalcpu", &num_physical_cores, &len, NULL, 0);
|
|
|
|
if (result == 0) {
|
|
|
|
return num_physical_cores;
|
|
|
|
}
|
|
|
|
result = sysctlbyname("hw.physicalcpu", &num_physical_cores, &len, NULL, 0);
|
|
|
|
if (result == 0) {
|
|
|
|
return num_physical_cores;
|
2023-03-17 18:47:35 +01:00
|
|
|
}
|
2023-04-30 20:41:35 +02:00
|
|
|
#elif defined(_WIN32)
|
|
|
|
//TODO: Implement
|
|
|
|
#endif
|
|
|
|
unsigned int n_threads = std::thread::hardware_concurrency();
|
|
|
|
return n_threads > 0 ? (n_threads <= 4 ? n_threads : n_threads / 2) : 4;
|
|
|
|
}
|
2023-03-17 18:47:35 +01:00
|
|
|
|
2023-05-04 14:08:25 +02:00
|
|
|
void process_escapes(std::string& input) {
|
|
|
|
std::size_t input_len = input.length();
|
|
|
|
std::size_t output_idx = 0;
|
2023-05-03 03:46:20 +02:00
|
|
|
|
2023-05-04 14:08:25 +02:00
|
|
|
for (std::size_t input_idx = 0; input_idx < input_len; ++input_idx) {
|
|
|
|
if (input[input_idx] == '\\' && input_idx + 1 < input_len) {
|
|
|
|
switch (input[++input_idx]) {
|
|
|
|
case 'n': input[output_idx++] = '\n'; break;
|
|
|
|
case 'r': input[output_idx++] = '\r'; break;
|
|
|
|
case 't': input[output_idx++] = '\t'; break;
|
|
|
|
case '\'': input[output_idx++] = '\''; break;
|
|
|
|
case '\"': input[output_idx++] = '\"'; break;
|
|
|
|
case '\\': input[output_idx++] = '\\'; break;
|
|
|
|
default: input[output_idx++] = '\\';
|
|
|
|
input[output_idx++] = input[input_idx]; break;
|
2023-05-03 03:46:20 +02:00
|
|
|
}
|
2023-05-04 14:08:25 +02:00
|
|
|
} else {
|
|
|
|
input[output_idx++] = input[input_idx];
|
2023-05-03 03:46:20 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-05-04 14:08:25 +02:00
|
|
|
input.resize(output_idx);
|
2023-05-03 03:46:20 +02:00
|
|
|
}
|
|
|
|
|
2023-04-30 20:41:35 +02:00
|
|
|
bool gpt_params_parse(int argc, char ** argv, gpt_params & params) {
|
2023-03-23 18:54:28 +01:00
|
|
|
bool invalid_param = false;
|
|
|
|
std::string arg;
|
2023-04-02 04:41:12 +02:00
|
|
|
gpt_params default_params;
|
2023-05-12 16:34:55 +02:00
|
|
|
const std::string arg_prefix = "--";
|
2023-04-02 04:41:12 +02:00
|
|
|
|
2023-03-10 19:40:58 +01:00
|
|
|
for (int i = 1; i < argc; i++) {
|
2023-03-23 18:54:28 +01:00
|
|
|
arg = argv[i];
|
2023-05-12 16:34:55 +02:00
|
|
|
if (arg.compare(0, arg_prefix.size(), arg_prefix) == 0) {
|
|
|
|
std::replace(arg.begin(), arg.end(), '_', '-');
|
|
|
|
}
|
2023-03-10 19:40:58 +01:00
|
|
|
|
|
|
|
if (arg == "-s" || arg == "--seed") {
|
2023-04-14 21:58:43 +02:00
|
|
|
if (++i >= argc) {
|
2023-03-23 18:54:28 +01:00
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-06-29 15:15:15 +02:00
|
|
|
params.seed = std::stoul(argv[i]);
|
2023-03-10 19:40:58 +01:00
|
|
|
} else if (arg == "-t" || arg == "--threads") {
|
2023-04-14 21:58:43 +02:00
|
|
|
if (++i >= argc) {
|
2023-03-23 18:54:28 +01:00
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-04-14 21:58:43 +02:00
|
|
|
params.n_threads = std::stoi(argv[i]);
|
2023-07-23 15:33:02 +02:00
|
|
|
if (params.n_threads <= 0) {
|
|
|
|
params.n_threads = std::thread::hardware_concurrency();
|
|
|
|
}
|
2023-03-10 19:40:58 +01:00
|
|
|
} else if (arg == "-p" || arg == "--prompt") {
|
2023-04-14 21:58:43 +02:00
|
|
|
if (++i >= argc) {
|
2023-03-23 18:54:28 +01:00
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-05-04 14:08:25 +02:00
|
|
|
params.prompt = argv[i];
|
2023-08-28 17:59:39 +02:00
|
|
|
} else if (arg == "-e" || arg == "--escape") {
|
|
|
|
params.escape = true;
|
2023-05-10 17:37:14 +02:00
|
|
|
} else if (arg == "--prompt-cache") {
|
2023-04-28 17:59:37 +02:00
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-05-10 17:37:14 +02:00
|
|
|
params.path_prompt_cache = argv[i];
|
|
|
|
} else if (arg == "--prompt-cache-all") {
|
|
|
|
params.prompt_cache_all = true;
|
2023-06-07 04:10:17 +02:00
|
|
|
} else if (arg == "--prompt-cache-ro") {
|
|
|
|
params.prompt_cache_ro = true;
|
2023-03-12 21:28:36 +01:00
|
|
|
} else if (arg == "-f" || arg == "--file") {
|
2023-04-14 21:58:43 +02:00
|
|
|
if (++i >= argc) {
|
2023-03-23 18:54:28 +01:00
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-04-14 21:58:43 +02:00
|
|
|
std::ifstream file(argv[i]);
|
2023-03-31 20:03:48 +02:00
|
|
|
if (!file) {
|
2023-04-14 21:58:43 +02:00
|
|
|
fprintf(stderr, "error: failed to open file '%s'\n", argv[i]);
|
2023-03-31 20:03:48 +02:00
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-03-19 17:37:02 +01:00
|
|
|
std::copy(std::istreambuf_iterator<char>(file), std::istreambuf_iterator<char>(), back_inserter(params.prompt));
|
2023-03-19 18:04:44 +01:00
|
|
|
if (params.prompt.back() == '\n') {
|
|
|
|
params.prompt.pop_back();
|
|
|
|
}
|
2023-05-12 16:34:55 +02:00
|
|
|
} else if (arg == "-n" || arg == "--n-predict") {
|
2023-04-14 21:58:43 +02:00
|
|
|
if (++i >= argc) {
|
2023-03-23 18:54:28 +01:00
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-04-14 21:58:43 +02:00
|
|
|
params.n_predict = std::stoi(argv[i]);
|
2023-05-12 16:34:55 +02:00
|
|
|
} else if (arg == "--top-k") {
|
2023-04-14 21:58:43 +02:00
|
|
|
if (++i >= argc) {
|
2023-03-23 18:54:28 +01:00
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-04-14 21:58:43 +02:00
|
|
|
params.top_k = std::stoi(argv[i]);
|
2023-05-12 16:34:55 +02:00
|
|
|
} else if (arg == "-c" || arg == "--ctx-size") {
|
2023-04-14 21:58:43 +02:00
|
|
|
if (++i >= argc) {
|
2023-03-23 18:54:28 +01:00
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-04-14 21:58:43 +02:00
|
|
|
params.n_ctx = std::stoi(argv[i]);
|
llama : add custom RoPE (#2054)
* Implement customizable RoPE
The original RoPE has pre-defined parameters
theta_i = 10000^(−2(i−1)/d), for i in [1, 2, ..., d/2]
Our customizable RoPE, ggml_rope_custom_inplace, uses
theta_i = scale * base^(−2(i−1)/d), for i in [1, 2, ..., d/2]
with the default matches the original
scale = 1.0
base = 10000
The new command line arguments
--rope-freq-base
--rope-freq-scale
set the two new RoPE parameter.
Recent researches show changing these two parameters extends the context limit with minimal loss.
1. Extending Context to 8K
kaiokendev
https://kaiokendev.github.io/til#extending-context-to-8k
2. Extending Context Window of Large Language Models via Positional Interpolation
Shouyuan Chen, Sherman Wong, Liangjian Chen, Yuandong Tian
https://arxiv.org/abs/2306.15595
3. NTK-Aware Scaled RoPE allows LLaMA models to have extended (8k+) context size without any fine-tuning and minimal perplexity degradation.
https://www.reddit.com/user/bloc97
https://www.reddit.com/r/LocalLLaMA/comments/14lz7j5/ntkaware_scaled_rope_allows_llama_models_to_have/
For the bold, try adding the following command line parameters to your favorite model:
-c 16384 --rope-freq-base 80000 --rope-freq-scale 0.5
* ggml-metal: fix custom rope
* common: fix argument names in help
* llama: increase MEM_REQ_EVAL for MODEL_3B
It avoids crashing for quantized weights on CPU.
Better ways to calculate the required buffer size would be better.
* llama: make MEM_REQ_EVAL depend on n_ctx
* server: use proper Content-Type in curl examples
Without the header Content-Type: application/json, curl will POST with
Content-Type: application/x-www-form-urlencoded
Though our simple server doesn't care, the httplib.h used has a limit
with CPPHTTPLIB_FORM_URL_ENCODED_PAYLOAD_MAX_LENGTH 8192
With Content-Type: application/json, we can send large json data.
* style : minor fixes, mostly indentations
* ggml : fix asserts
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-07-15 12:34:16 +02:00
|
|
|
} else if (arg == "--rope-freq-base") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.rope_freq_base = std::stof(argv[i]);
|
|
|
|
} else if (arg == "--rope-freq-scale") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.rope_freq_scale = std::stof(argv[i]);
|
2023-08-07 19:07:19 +02:00
|
|
|
} else if (arg == "--rope-scale") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.rope_freq_scale = 1.0f/std::stof(argv[i]);
|
2023-05-12 16:34:55 +02:00
|
|
|
} else if (arg == "--memory-f32") {
|
2023-03-24 22:17:37 +01:00
|
|
|
params.memory_f16 = false;
|
2023-05-12 16:34:55 +02:00
|
|
|
} else if (arg == "--top-p") {
|
2023-04-14 21:58:43 +02:00
|
|
|
if (++i >= argc) {
|
2023-03-23 18:54:28 +01:00
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-04-14 21:58:43 +02:00
|
|
|
params.top_p = std::stof(argv[i]);
|
2023-03-10 19:40:58 +01:00
|
|
|
} else if (arg == "--temp") {
|
2023-04-14 21:58:43 +02:00
|
|
|
if (++i >= argc) {
|
2023-03-23 18:54:28 +01:00
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-04-14 21:58:43 +02:00
|
|
|
params.temp = std::stof(argv[i]);
|
llama : new sampling algorithms (#1126)
* Sample interface, new samplers.
New samplers:
- locally typical sampling
- tail free sampling
- frequency and presence penalty
- mirostat
Ignore EOS fix: -inf should be used.
* mirostat
* Added --logit-bias and --no-penalize-nl, removed std::span
* Use C++11, clarify llama API documentation, rename Mirostat parameters to --mirostat_lr and --mirostat_ent, add temperature sampling for Mirostat, simplify Mirostat sampling API parameters (removed N and *k)
Use C++11, clarify llama API documentation, rename Mirostat parameters to --mirostat_lr and --mirostat_ent, add temperature sampling for Mirostat, simplify Mirostat sampling API parameters (removed N and *k)
* Save and load example adjust
* Tests
* Windows build fix
* Windows test fix
2023-04-29 07:34:41 +02:00
|
|
|
} else if (arg == "--tfs") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.tfs_z = std::stof(argv[i]);
|
|
|
|
} else if (arg == "--typical") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.typical_p = std::stof(argv[i]);
|
2023-05-12 16:34:55 +02:00
|
|
|
} else if (arg == "--repeat-last-n") {
|
2023-04-14 21:58:43 +02:00
|
|
|
if (++i >= argc) {
|
2023-03-23 18:54:28 +01:00
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-04-14 21:58:43 +02:00
|
|
|
params.repeat_last_n = std::stoi(argv[i]);
|
2023-05-12 16:34:55 +02:00
|
|
|
} else if (arg == "--repeat-penalty") {
|
2023-04-14 21:58:43 +02:00
|
|
|
if (++i >= argc) {
|
2023-03-23 18:54:28 +01:00
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-04-14 21:58:43 +02:00
|
|
|
params.repeat_penalty = std::stof(argv[i]);
|
2023-05-12 16:34:55 +02:00
|
|
|
} else if (arg == "--frequency-penalty") {
|
llama : new sampling algorithms (#1126)
* Sample interface, new samplers.
New samplers:
- locally typical sampling
- tail free sampling
- frequency and presence penalty
- mirostat
Ignore EOS fix: -inf should be used.
* mirostat
* Added --logit-bias and --no-penalize-nl, removed std::span
* Use C++11, clarify llama API documentation, rename Mirostat parameters to --mirostat_lr and --mirostat_ent, add temperature sampling for Mirostat, simplify Mirostat sampling API parameters (removed N and *k)
Use C++11, clarify llama API documentation, rename Mirostat parameters to --mirostat_lr and --mirostat_ent, add temperature sampling for Mirostat, simplify Mirostat sampling API parameters (removed N and *k)
* Save and load example adjust
* Tests
* Windows build fix
* Windows test fix
2023-04-29 07:34:41 +02:00
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.frequency_penalty = std::stof(argv[i]);
|
2023-05-12 16:34:55 +02:00
|
|
|
} else if (arg == "--presence-penalty") {
|
llama : new sampling algorithms (#1126)
* Sample interface, new samplers.
New samplers:
- locally typical sampling
- tail free sampling
- frequency and presence penalty
- mirostat
Ignore EOS fix: -inf should be used.
* mirostat
* Added --logit-bias and --no-penalize-nl, removed std::span
* Use C++11, clarify llama API documentation, rename Mirostat parameters to --mirostat_lr and --mirostat_ent, add temperature sampling for Mirostat, simplify Mirostat sampling API parameters (removed N and *k)
Use C++11, clarify llama API documentation, rename Mirostat parameters to --mirostat_lr and --mirostat_ent, add temperature sampling for Mirostat, simplify Mirostat sampling API parameters (removed N and *k)
* Save and load example adjust
* Tests
* Windows build fix
* Windows test fix
2023-04-29 07:34:41 +02:00
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.presence_penalty = std::stof(argv[i]);
|
|
|
|
} else if (arg == "--mirostat") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.mirostat = std::stoi(argv[i]);
|
2023-05-12 16:34:55 +02:00
|
|
|
} else if (arg == "--mirostat-lr") {
|
llama : new sampling algorithms (#1126)
* Sample interface, new samplers.
New samplers:
- locally typical sampling
- tail free sampling
- frequency and presence penalty
- mirostat
Ignore EOS fix: -inf should be used.
* mirostat
* Added --logit-bias and --no-penalize-nl, removed std::span
* Use C++11, clarify llama API documentation, rename Mirostat parameters to --mirostat_lr and --mirostat_ent, add temperature sampling for Mirostat, simplify Mirostat sampling API parameters (removed N and *k)
Use C++11, clarify llama API documentation, rename Mirostat parameters to --mirostat_lr and --mirostat_ent, add temperature sampling for Mirostat, simplify Mirostat sampling API parameters (removed N and *k)
* Save and load example adjust
* Tests
* Windows build fix
* Windows test fix
2023-04-29 07:34:41 +02:00
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.mirostat_eta = std::stof(argv[i]);
|
2023-05-12 16:34:55 +02:00
|
|
|
} else if (arg == "--mirostat-ent") {
|
llama : new sampling algorithms (#1126)
* Sample interface, new samplers.
New samplers:
- locally typical sampling
- tail free sampling
- frequency and presence penalty
- mirostat
Ignore EOS fix: -inf should be used.
* mirostat
* Added --logit-bias and --no-penalize-nl, removed std::span
* Use C++11, clarify llama API documentation, rename Mirostat parameters to --mirostat_lr and --mirostat_ent, add temperature sampling for Mirostat, simplify Mirostat sampling API parameters (removed N and *k)
Use C++11, clarify llama API documentation, rename Mirostat parameters to --mirostat_lr and --mirostat_ent, add temperature sampling for Mirostat, simplify Mirostat sampling API parameters (removed N and *k)
* Save and load example adjust
* Tests
* Windows build fix
* Windows test fix
2023-04-29 07:34:41 +02:00
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.mirostat_tau = std::stof(argv[i]);
|
2023-07-11 18:18:43 +02:00
|
|
|
} else if (arg == "--cfg-negative-prompt") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.cfg_negative_prompt = argv[i];
|
2023-08-17 15:29:44 +02:00
|
|
|
} else if (arg == "--cfg-negative-prompt-file") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
std::ifstream file(argv[i]);
|
|
|
|
if (!file) {
|
|
|
|
fprintf(stderr, "error: failed to open file '%s'\n", argv[i]);
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
std::copy(std::istreambuf_iterator<char>(file), std::istreambuf_iterator<char>(), back_inserter(params.cfg_negative_prompt));
|
|
|
|
if (params.cfg_negative_prompt.back() == '\n') {
|
|
|
|
params.cfg_negative_prompt.pop_back();
|
|
|
|
}
|
2023-07-11 18:18:43 +02:00
|
|
|
} else if (arg == "--cfg-scale") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.cfg_scale = std::stof(argv[i]);
|
2023-05-12 16:34:55 +02:00
|
|
|
} else if (arg == "-b" || arg == "--batch-size") {
|
2023-04-14 21:58:43 +02:00
|
|
|
if (++i >= argc) {
|
2023-03-23 18:54:28 +01:00
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-04-14 21:58:43 +02:00
|
|
|
params.n_batch = std::stoi(argv[i]);
|
2023-03-25 20:36:22 +01:00
|
|
|
} else if (arg == "--keep") {
|
2023-04-14 21:58:43 +02:00
|
|
|
if (++i >= argc) {
|
2023-03-25 20:36:22 +01:00
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-04-14 21:58:43 +02:00
|
|
|
params.n_keep = std::stoi(argv[i]);
|
2023-07-18 13:24:43 +02:00
|
|
|
} else if (arg == "--chunks") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.n_chunks = std::stoi(argv[i]);
|
2023-03-10 19:40:58 +01:00
|
|
|
} else if (arg == "-m" || arg == "--model") {
|
2023-04-14 21:58:43 +02:00
|
|
|
if (++i >= argc) {
|
2023-03-23 18:54:28 +01:00
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-04-14 21:58:43 +02:00
|
|
|
params.model = argv[i];
|
2023-05-28 19:14:24 +02:00
|
|
|
} else if (arg == "-a" || arg == "--alias") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.model_alias = argv[i];
|
2023-04-17 17:28:55 +02:00
|
|
|
} else if (arg == "--lora") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.lora_adapter = argv[i];
|
2023-07-13 15:58:25 +02:00
|
|
|
params.use_mmap = false;
|
2023-04-17 17:28:55 +02:00
|
|
|
} else if (arg == "--lora-base") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.lora_base = argv[i];
|
2023-03-12 22:13:28 +01:00
|
|
|
} else if (arg == "-i" || arg == "--interactive") {
|
|
|
|
params.interactive = true;
|
2023-03-24 16:05:13 +01:00
|
|
|
} else if (arg == "--embedding") {
|
|
|
|
params.embedding = true;
|
2023-03-22 18:16:35 +01:00
|
|
|
} else if (arg == "--interactive-first") {
|
2023-04-24 17:45:32 +02:00
|
|
|
params.interactive_first = true;
|
2023-03-19 17:37:02 +01:00
|
|
|
} else if (arg == "-ins" || arg == "--instruct") {
|
2023-04-14 21:58:43 +02:00
|
|
|
params.instruct = true;
|
2023-05-09 04:45:48 +02:00
|
|
|
} else if (arg == "--multiline-input") {
|
|
|
|
params.multiline_input = true;
|
2023-08-04 17:20:12 +02:00
|
|
|
} else if (arg == "--simple-io") {
|
|
|
|
params.simple_io = true;
|
2023-03-12 22:13:28 +01:00
|
|
|
} else if (arg == "--color") {
|
|
|
|
params.use_color = true;
|
2023-03-24 16:19:05 +01:00
|
|
|
} else if (arg == "--mlock") {
|
|
|
|
params.use_mlock = true;
|
2023-05-13 15:38:36 +02:00
|
|
|
} else if (arg == "--gpu-layers" || arg == "-ngl" || arg == "--n-gpu-layers") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-05-28 19:48:57 +02:00
|
|
|
#ifdef LLAMA_SUPPORTS_GPU_OFFLOAD
|
2023-05-13 15:38:36 +02:00
|
|
|
params.n_gpu_layers = std::stoi(argv[i]);
|
2023-05-28 19:48:57 +02:00
|
|
|
#else
|
|
|
|
fprintf(stderr, "warning: not compiled with GPU offload support, --n-gpu-layers option will be ignored\n");
|
|
|
|
fprintf(stderr, "warning: see main README.md for information on enabling GPU BLAS support\n");
|
|
|
|
#endif
|
2023-06-06 21:33:23 +02:00
|
|
|
} else if (arg == "--main-gpu" || arg == "-mg") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
#ifdef GGML_USE_CUBLAS
|
|
|
|
params.main_gpu = std::stoi(argv[i]);
|
|
|
|
#else
|
2023-07-31 15:44:35 +02:00
|
|
|
fprintf(stderr, "warning: llama.cpp was compiled without cuBLAS. It is not possible to set a main GPU.\n");
|
2023-06-06 21:33:23 +02:00
|
|
|
#endif
|
|
|
|
} else if (arg == "--tensor-split" || arg == "-ts") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
#ifdef GGML_USE_CUBLAS
|
|
|
|
std::string arg_next = argv[i];
|
|
|
|
|
|
|
|
// split string by , and /
|
|
|
|
const std::regex regex{R"([,/]+)"};
|
|
|
|
std::sregex_token_iterator it{arg_next.begin(), arg_next.end(), regex, -1};
|
|
|
|
std::vector<std::string> split_arg{it, {}};
|
|
|
|
GGML_ASSERT(split_arg.size() <= LLAMA_MAX_DEVICES);
|
|
|
|
|
|
|
|
for (size_t i = 0; i < LLAMA_MAX_DEVICES; ++i) {
|
|
|
|
if (i < split_arg.size()) {
|
|
|
|
params.tensor_split[i] = std::stof(split_arg[i]);
|
|
|
|
} else {
|
|
|
|
params.tensor_split[i] = 0.0f;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
#else
|
2023-07-31 15:44:35 +02:00
|
|
|
fprintf(stderr, "warning: llama.cpp was compiled without cuBLAS. It is not possible to set a tensor split.\n");
|
|
|
|
#endif // GGML_USE_CUBLAS
|
2023-08-22 22:47:05 +02:00
|
|
|
} else if (arg == "--no-mul-mat-q" || arg == "-nommq") {
|
2023-07-31 15:44:35 +02:00
|
|
|
#ifdef GGML_USE_CUBLAS
|
2023-08-22 22:47:05 +02:00
|
|
|
params.mul_mat_q = false;
|
2023-07-31 15:44:35 +02:00
|
|
|
#else
|
2023-08-22 22:47:05 +02:00
|
|
|
fprintf(stderr, "warning: llama.cpp was compiled without cuBLAS. Disabling mul_mat_q kernels has no effect.\n");
|
2023-06-14 19:47:19 +02:00
|
|
|
#endif // GGML_USE_CUBLAS
|
|
|
|
} else if (arg == "--low-vram" || arg == "-lv") {
|
|
|
|
#ifdef GGML_USE_CUBLAS
|
|
|
|
params.low_vram = true;
|
|
|
|
#else
|
2023-07-31 15:44:35 +02:00
|
|
|
fprintf(stderr, "warning: llama.cpp was compiled without cuBLAS. It is not possible to set lower vram usage.\n");
|
2023-06-06 21:33:23 +02:00
|
|
|
#endif // GGML_USE_CUBLAS
|
Rewrite loading code to try to satisfy everyone:
- Support all three formats (ggml, ggmf, ggjt). (However, I didn't
include the hack needed to support GPT4All files without conversion.
Those can still be used after converting them with convert.py from my
other PR.)
- Support both mmap and read (mmap is used by default, but can be
disabled with `--no-mmap`, and is automatically disabled for pre-ggjt
files or on platforms where mmap is not supported).
- Support multi-file models like before, but automatically determine the
number of parts rather than requiring `--n_parts`.
- Improve validation and error checking.
- Stop using the per-file type field (f16) entirely in favor of just
relying on the per-tensor type/size fields. This has no immediate
benefit, but makes it easier to experiment with different formats, and
should make it easier to support the new GPTQ-for-LLaMa models in the
future (I have some work in progress on that front).
- Support VirtualLock on Windows (using the same `--mlock` option as on
Unix).
- Indicate loading progress when using mmap + mlock. (Which led me
to the interesting observation that on my Linux machine, with a
warm file cache, mlock actually takes some time, whereas mmap
without mlock starts almost instantly...)
- To help implement this, move mlock support from ggml to the
loading code.
- madvise/PrefetchVirtualMemory support (based on #740)
- Switch from ifstream to the `fopen` family of functions to avoid
unnecessary copying and, when mmap is enabled, allow reusing the same
file descriptor for both metadata reads and mmap (whereas the existing
implementation opens the file a second time to mmap).
- Quantization now produces a single-file output even with multi-file
inputs (not really a feature as much as 'it was easier this way').
Implementation notes:
I tried to factor the code into more discrete pieces than before.
Regarding code style: I tried to follow the code style, but I'm naughty
and used a few advanced C++ features repeatedly:
- Destructors to make it easier to ensure everything gets cleaned up.
- Exceptions. I don't even usually use exceptions when writing C++, and
I can remove them if desired... but here they make the loading code
much more succinct while still properly handling a variety of errors,
ranging from API calls failing to integer overflow and allocation
failure. The exceptions are converted to error codes at the
API boundary.)
Co-authored-by: Pavol Rusnak <pavol@rusnak.io> (for the bit I copied from #740)
2023-04-08 21:24:37 +02:00
|
|
|
} else if (arg == "--no-mmap") {
|
|
|
|
params.use_mmap = false;
|
2023-03-24 22:17:37 +01:00
|
|
|
} else if (arg == "--mtest") {
|
|
|
|
params.mem_test = true;
|
2023-06-26 19:57:59 +02:00
|
|
|
} else if (arg == "--numa") {
|
|
|
|
params.numa = true;
|
2023-06-04 22:34:30 +02:00
|
|
|
} else if (arg == "--export") {
|
|
|
|
params.export_cgraph = true;
|
2023-03-25 20:36:22 +01:00
|
|
|
} else if (arg == "--verbose-prompt") {
|
2023-03-25 16:16:50 +01:00
|
|
|
params.verbose_prompt = true;
|
2023-03-12 22:13:28 +01:00
|
|
|
} else if (arg == "-r" || arg == "--reverse-prompt") {
|
2023-04-14 21:58:43 +02:00
|
|
|
if (++i >= argc) {
|
2023-04-14 17:19:17 +02:00
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-04-14 21:58:43 +02:00
|
|
|
params.antiprompt.push_back(argv[i]);
|
2023-08-28 17:59:39 +02:00
|
|
|
} else if (arg == "-ld" || arg == "--logdir") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.logdir = argv[i];
|
|
|
|
|
|
|
|
if (params.logdir.back() != DIRECTORY_SEPARATOR) {
|
|
|
|
params.logdir += DIRECTORY_SEPARATOR;
|
|
|
|
}
|
2023-03-21 17:27:42 +01:00
|
|
|
} else if (arg == "--perplexity") {
|
|
|
|
params.perplexity = true;
|
2023-08-23 11:56:42 +02:00
|
|
|
} else if (arg == "--ppl-stride") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.ppl_stride = std::stoi(argv[i]);
|
|
|
|
} else if (arg == "--ppl-output-type") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.ppl_output_type = std::stoi(argv[i]);
|
2023-07-28 20:25:36 +02:00
|
|
|
} else if (arg == "--hellaswag") {
|
|
|
|
params.hellaswag = true;
|
|
|
|
} else if (arg == "--hellaswag-tasks") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.hellaswag_tasks = std::stoi(argv[i]);
|
2023-03-19 19:22:48 +01:00
|
|
|
} else if (arg == "--ignore-eos") {
|
2023-08-21 22:07:43 +02:00
|
|
|
params.ignore_eos = true;
|
llama : new sampling algorithms (#1126)
* Sample interface, new samplers.
New samplers:
- locally typical sampling
- tail free sampling
- frequency and presence penalty
- mirostat
Ignore EOS fix: -inf should be used.
* mirostat
* Added --logit-bias and --no-penalize-nl, removed std::span
* Use C++11, clarify llama API documentation, rename Mirostat parameters to --mirostat_lr and --mirostat_ent, add temperature sampling for Mirostat, simplify Mirostat sampling API parameters (removed N and *k)
Use C++11, clarify llama API documentation, rename Mirostat parameters to --mirostat_lr and --mirostat_ent, add temperature sampling for Mirostat, simplify Mirostat sampling API parameters (removed N and *k)
* Save and load example adjust
* Tests
* Windows build fix
* Windows test fix
2023-04-29 07:34:41 +02:00
|
|
|
} else if (arg == "--no-penalize-nl") {
|
|
|
|
params.penalize_nl = false;
|
|
|
|
} else if (arg == "-l" || arg == "--logit-bias") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
std::stringstream ss(argv[i]);
|
|
|
|
llama_token key;
|
|
|
|
char sign;
|
|
|
|
std::string value_str;
|
|
|
|
try {
|
|
|
|
if (ss >> key && ss >> sign && std::getline(ss, value_str) && (sign == '+' || sign == '-')) {
|
|
|
|
params.logit_bias[key] = std::stof(value_str) * ((sign == '-') ? -1.0f : 1.0f);
|
|
|
|
} else {
|
|
|
|
throw std::exception();
|
|
|
|
}
|
2023-06-16 20:23:53 +02:00
|
|
|
} catch (const std::exception&) {
|
llama : new sampling algorithms (#1126)
* Sample interface, new samplers.
New samplers:
- locally typical sampling
- tail free sampling
- frequency and presence penalty
- mirostat
Ignore EOS fix: -inf should be used.
* mirostat
* Added --logit-bias and --no-penalize-nl, removed std::span
* Use C++11, clarify llama API documentation, rename Mirostat parameters to --mirostat_lr and --mirostat_ent, add temperature sampling for Mirostat, simplify Mirostat sampling API parameters (removed N and *k)
Use C++11, clarify llama API documentation, rename Mirostat parameters to --mirostat_lr and --mirostat_ent, add temperature sampling for Mirostat, simplify Mirostat sampling API parameters (removed N and *k)
* Save and load example adjust
* Tests
* Windows build fix
* Windows test fix
2023-04-29 07:34:41 +02:00
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-03-10 19:40:58 +01:00
|
|
|
} else if (arg == "-h" || arg == "--help") {
|
2023-04-14 21:58:43 +02:00
|
|
|
gpt_print_usage(argc, argv, default_params);
|
2023-03-10 19:40:58 +01:00
|
|
|
exit(0);
|
2023-03-19 19:36:19 +01:00
|
|
|
} else if (arg == "--random-prompt") {
|
|
|
|
params.random_prompt = true;
|
2023-07-25 14:19:11 +02:00
|
|
|
} else if (arg == "--in-prefix-bos") {
|
|
|
|
params.input_prefix_bos = true;
|
2023-03-25 13:03:19 +01:00
|
|
|
} else if (arg == "--in-prefix") {
|
2023-04-14 21:58:43 +02:00
|
|
|
if (++i >= argc) {
|
2023-04-14 17:19:17 +02:00
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
2023-04-14 21:58:43 +02:00
|
|
|
params.input_prefix = argv[i];
|
2023-05-04 17:41:12 +02:00
|
|
|
} else if (arg == "--in-suffix") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.input_suffix = argv[i];
|
2023-07-24 05:58:10 +02:00
|
|
|
} else if (arg == "--grammar") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
params.grammar = argv[i];
|
|
|
|
} else if (arg == "--grammar-file") {
|
|
|
|
if (++i >= argc) {
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
std::ifstream file(argv[i]);
|
|
|
|
if (!file) {
|
|
|
|
fprintf(stderr, "error: failed to open file '%s'\n", argv[i]);
|
|
|
|
invalid_param = true;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
std::copy(
|
|
|
|
std::istreambuf_iterator<char>(file),
|
|
|
|
std::istreambuf_iterator<char>(),
|
|
|
|
std::back_inserter(params.grammar)
|
|
|
|
);
|
2023-03-10 19:40:58 +01:00
|
|
|
} else {
|
|
|
|
fprintf(stderr, "error: unknown argument: %s\n", arg.c_str());
|
2023-04-14 21:58:43 +02:00
|
|
|
gpt_print_usage(argc, argv, default_params);
|
2023-03-23 18:54:28 +01:00
|
|
|
exit(1);
|
2023-03-10 19:40:58 +01:00
|
|
|
}
|
|
|
|
}
|
2023-03-23 18:54:28 +01:00
|
|
|
if (invalid_param) {
|
|
|
|
fprintf(stderr, "error: invalid parameter for argument: %s\n", arg.c_str());
|
2023-04-14 21:58:43 +02:00
|
|
|
gpt_print_usage(argc, argv, default_params);
|
2023-03-23 18:54:28 +01:00
|
|
|
exit(1);
|
|
|
|
}
|
2023-05-10 17:37:14 +02:00
|
|
|
if (params.prompt_cache_all &&
|
|
|
|
(params.interactive || params.interactive_first ||
|
2023-05-19 19:24:59 +02:00
|
|
|
params.instruct)) {
|
2023-05-10 17:37:14 +02:00
|
|
|
fprintf(stderr, "error: --prompt-cache-all not supported in interactive mode yet\n");
|
|
|
|
gpt_print_usage(argc, argv, default_params);
|
|
|
|
exit(1);
|
|
|
|
}
|
2023-06-15 19:06:46 +02:00
|
|
|
|
2023-08-28 17:59:39 +02:00
|
|
|
if (params.escape) {
|
2023-05-04 14:08:25 +02:00
|
|
|
process_escapes(params.prompt);
|
2023-07-09 10:56:18 +02:00
|
|
|
process_escapes(params.input_prefix);
|
|
|
|
process_escapes(params.input_suffix);
|
2023-05-04 14:08:25 +02:00
|
|
|
}
|
2023-03-10 19:40:58 +01:00
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2023-04-14 21:58:43 +02:00
|
|
|
void gpt_print_usage(int /*argc*/, char ** argv, const gpt_params & params) {
|
2023-07-23 13:59:48 +02:00
|
|
|
fprintf(stdout, "usage: %s [options]\n", argv[0]);
|
|
|
|
fprintf(stdout, "\n");
|
|
|
|
fprintf(stdout, "options:\n");
|
|
|
|
fprintf(stdout, " -h, --help show this help message and exit\n");
|
|
|
|
fprintf(stdout, " -i, --interactive run in interactive mode\n");
|
|
|
|
fprintf(stdout, " --interactive-first run in interactive mode and wait for input right away\n");
|
|
|
|
fprintf(stdout, " -ins, --instruct run in instruction mode (use with Alpaca models)\n");
|
|
|
|
fprintf(stdout, " --multiline-input allows you to write or paste multiple lines without ending each in '\\'\n");
|
|
|
|
fprintf(stdout, " -r PROMPT, --reverse-prompt PROMPT\n");
|
|
|
|
fprintf(stdout, " halt generation at PROMPT, return control in interactive mode\n");
|
|
|
|
fprintf(stdout, " (can be specified more than once for multiple prompts).\n");
|
|
|
|
fprintf(stdout, " --color colorise output to distinguish prompt and user input from generations\n");
|
|
|
|
fprintf(stdout, " -s SEED, --seed SEED RNG seed (default: -1, use random seed for < 0)\n");
|
|
|
|
fprintf(stdout, " -t N, --threads N number of threads to use during computation (default: %d)\n", params.n_threads);
|
|
|
|
fprintf(stdout, " -p PROMPT, --prompt PROMPT\n");
|
|
|
|
fprintf(stdout, " prompt to start generation with (default: empty)\n");
|
2023-08-28 17:59:39 +02:00
|
|
|
fprintf(stdout, " -e, --escape process prompt escapes sequences (\\n, \\r, \\t, \\', \\\", \\\\)\n");
|
2023-07-23 13:59:48 +02:00
|
|
|
fprintf(stdout, " --prompt-cache FNAME file to cache prompt state for faster startup (default: none)\n");
|
|
|
|
fprintf(stdout, " --prompt-cache-all if specified, saves user input and generations to cache as well.\n");
|
|
|
|
fprintf(stdout, " not supported with --interactive or other interactive options\n");
|
|
|
|
fprintf(stdout, " --prompt-cache-ro if specified, uses the prompt cache but does not update it.\n");
|
|
|
|
fprintf(stdout, " --random-prompt start with a randomized prompt.\n");
|
2023-07-25 14:19:11 +02:00
|
|
|
fprintf(stdout, " --in-prefix-bos prefix BOS to user inputs, preceding the `--in-prefix` string\n");
|
2023-07-23 13:59:48 +02:00
|
|
|
fprintf(stdout, " --in-prefix STRING string to prefix user inputs with (default: empty)\n");
|
|
|
|
fprintf(stdout, " --in-suffix STRING string to suffix after user inputs with (default: empty)\n");
|
|
|
|
fprintf(stdout, " -f FNAME, --file FNAME\n");
|
|
|
|
fprintf(stdout, " prompt file to start generation.\n");
|
2023-08-10 16:28:27 +02:00
|
|
|
fprintf(stdout, " -n N, --n-predict N number of tokens to predict (default: %d, -1 = infinity, -2 = until context filled)\n", params.n_predict);
|
2023-07-23 14:09:47 +02:00
|
|
|
fprintf(stdout, " -c N, --ctx-size N size of the prompt context (default: %d)\n", params.n_ctx);
|
|
|
|
fprintf(stdout, " -b N, --batch-size N batch size for prompt processing (default: %d)\n", params.n_batch);
|
2023-07-23 13:59:48 +02:00
|
|
|
fprintf(stdout, " --top-k N top-k sampling (default: %d, 0 = disabled)\n", params.top_k);
|
|
|
|
fprintf(stdout, " --top-p N top-p sampling (default: %.1f, 1.0 = disabled)\n", (double)params.top_p);
|
|
|
|
fprintf(stdout, " --tfs N tail free sampling, parameter z (default: %.1f, 1.0 = disabled)\n", (double)params.tfs_z);
|
|
|
|
fprintf(stdout, " --typical N locally typical sampling, parameter p (default: %.1f, 1.0 = disabled)\n", (double)params.typical_p);
|
|
|
|
fprintf(stdout, " --repeat-last-n N last n tokens to consider for penalize (default: %d, 0 = disabled, -1 = ctx_size)\n", params.repeat_last_n);
|
|
|
|
fprintf(stdout, " --repeat-penalty N penalize repeat sequence of tokens (default: %.1f, 1.0 = disabled)\n", (double)params.repeat_penalty);
|
|
|
|
fprintf(stdout, " --presence-penalty N repeat alpha presence penalty (default: %.1f, 0.0 = disabled)\n", (double)params.presence_penalty);
|
|
|
|
fprintf(stdout, " --frequency-penalty N repeat alpha frequency penalty (default: %.1f, 0.0 = disabled)\n", (double)params.frequency_penalty);
|
|
|
|
fprintf(stdout, " --mirostat N use Mirostat sampling.\n");
|
|
|
|
fprintf(stdout, " Top K, Nucleus, Tail Free and Locally Typical samplers are ignored if used.\n");
|
|
|
|
fprintf(stdout, " (default: %d, 0 = disabled, 1 = Mirostat, 2 = Mirostat 2.0)\n", params.mirostat);
|
|
|
|
fprintf(stdout, " --mirostat-lr N Mirostat learning rate, parameter eta (default: %.1f)\n", (double)params.mirostat_eta);
|
|
|
|
fprintf(stdout, " --mirostat-ent N Mirostat target entropy, parameter tau (default: %.1f)\n", (double)params.mirostat_tau);
|
|
|
|
fprintf(stdout, " -l TOKEN_ID(+/-)BIAS, --logit-bias TOKEN_ID(+/-)BIAS\n");
|
|
|
|
fprintf(stdout, " modifies the likelihood of token appearing in the completion,\n");
|
|
|
|
fprintf(stdout, " i.e. `--logit-bias 15043+1` to increase likelihood of token ' Hello',\n");
|
|
|
|
fprintf(stdout, " or `--logit-bias 15043-1` to decrease likelihood of token ' Hello'\n");
|
2023-07-24 05:58:10 +02:00
|
|
|
fprintf(stdout, " --grammar GRAMMAR BNF-like grammar to constrain generations (see samples in grammars/ dir)\n");
|
|
|
|
fprintf(stdout, " --grammar-file FNAME file to read grammar from\n");
|
2023-08-17 15:29:44 +02:00
|
|
|
fprintf(stdout, " --cfg-negative-prompt PROMPT\n");
|
2023-07-23 13:59:48 +02:00
|
|
|
fprintf(stdout, " negative prompt to use for guidance. (default: empty)\n");
|
2023-08-17 15:29:44 +02:00
|
|
|
fprintf(stdout, " --cfg-negative-prompt-file FNAME\n");
|
|
|
|
fprintf(stdout, " negative prompt file to use for guidance. (default: empty)\n");
|
2023-07-23 13:59:48 +02:00
|
|
|
fprintf(stdout, " --cfg-scale N strength of guidance (default: %f, 1.0 = disable)\n", params.cfg_scale);
|
2023-08-07 19:07:19 +02:00
|
|
|
fprintf(stdout, " --rope-scale N RoPE context linear scaling factor, inverse of --rope-freq-scale (default: %g)\n", 1.0f/params.rope_freq_scale);
|
|
|
|
fprintf(stdout, " --rope-freq-base N RoPE base frequency, used by NTK-aware scaling (default: %.1f)\n", params.rope_freq_base);
|
|
|
|
fprintf(stdout, " --rope-freq-scale N RoPE frequency linear scaling factor, inverse of --rope-scale (default: %g)\n", params.rope_freq_scale);
|
2023-07-23 13:59:48 +02:00
|
|
|
fprintf(stdout, " --ignore-eos ignore end of stream token and continue generating (implies --logit-bias 2-inf)\n");
|
|
|
|
fprintf(stdout, " --no-penalize-nl do not penalize newline token\n");
|
|
|
|
fprintf(stdout, " --memory-f32 use f32 instead of f16 for memory key+value (default: disabled)\n");
|
|
|
|
fprintf(stdout, " not recommended: doubles context memory required and no measurable increase in quality\n");
|
|
|
|
fprintf(stdout, " --temp N temperature (default: %.1f)\n", (double)params.temp);
|
|
|
|
fprintf(stdout, " --perplexity compute perplexity over each ctx window of the prompt\n");
|
2023-07-28 20:25:36 +02:00
|
|
|
fprintf(stdout, " --hellaswag compute HellaSwag score over random tasks from datafile supplied with -f\n");
|
2023-08-02 10:06:19 +02:00
|
|
|
fprintf(stdout, " --hellaswag-tasks N number of tasks to use when computing the HellaSwag score (default: %zu)\n", params.hellaswag_tasks);
|
2023-07-28 20:25:36 +02:00
|
|
|
fprintf(stdout, " --keep N number of tokens to keep from the initial prompt (default: %d, -1 = all)\n", params.n_keep);
|
2023-07-23 13:59:48 +02:00
|
|
|
fprintf(stdout, " --chunks N max number of chunks to process (default: %d, -1 = all)\n", params.n_chunks);
|
Rewrite loading code to try to satisfy everyone:
- Support all three formats (ggml, ggmf, ggjt). (However, I didn't
include the hack needed to support GPT4All files without conversion.
Those can still be used after converting them with convert.py from my
other PR.)
- Support both mmap and read (mmap is used by default, but can be
disabled with `--no-mmap`, and is automatically disabled for pre-ggjt
files or on platforms where mmap is not supported).
- Support multi-file models like before, but automatically determine the
number of parts rather than requiring `--n_parts`.
- Improve validation and error checking.
- Stop using the per-file type field (f16) entirely in favor of just
relying on the per-tensor type/size fields. This has no immediate
benefit, but makes it easier to experiment with different formats, and
should make it easier to support the new GPTQ-for-LLaMa models in the
future (I have some work in progress on that front).
- Support VirtualLock on Windows (using the same `--mlock` option as on
Unix).
- Indicate loading progress when using mmap + mlock. (Which led me
to the interesting observation that on my Linux machine, with a
warm file cache, mlock actually takes some time, whereas mmap
without mlock starts almost instantly...)
- To help implement this, move mlock support from ggml to the
loading code.
- madvise/PrefetchVirtualMemory support (based on #740)
- Switch from ifstream to the `fopen` family of functions to avoid
unnecessary copying and, when mmap is enabled, allow reusing the same
file descriptor for both metadata reads and mmap (whereas the existing
implementation opens the file a second time to mmap).
- Quantization now produces a single-file output even with multi-file
inputs (not really a feature as much as 'it was easier this way').
Implementation notes:
I tried to factor the code into more discrete pieces than before.
Regarding code style: I tried to follow the code style, but I'm naughty
and used a few advanced C++ features repeatedly:
- Destructors to make it easier to ensure everything gets cleaned up.
- Exceptions. I don't even usually use exceptions when writing C++, and
I can remove them if desired... but here they make the loading code
much more succinct while still properly handling a variety of errors,
ranging from API calls failing to integer overflow and allocation
failure. The exceptions are converted to error codes at the
API boundary.)
Co-authored-by: Pavol Rusnak <pavol@rusnak.io> (for the bit I copied from #740)
2023-04-08 21:24:37 +02:00
|
|
|
if (llama_mlock_supported()) {
|
2023-07-23 13:59:48 +02:00
|
|
|
fprintf(stdout, " --mlock force system to keep model in RAM rather than swapping or compressing\n");
|
2023-03-24 16:19:05 +01:00
|
|
|
}
|
Rewrite loading code to try to satisfy everyone:
- Support all three formats (ggml, ggmf, ggjt). (However, I didn't
include the hack needed to support GPT4All files without conversion.
Those can still be used after converting them with convert.py from my
other PR.)
- Support both mmap and read (mmap is used by default, but can be
disabled with `--no-mmap`, and is automatically disabled for pre-ggjt
files or on platforms where mmap is not supported).
- Support multi-file models like before, but automatically determine the
number of parts rather than requiring `--n_parts`.
- Improve validation and error checking.
- Stop using the per-file type field (f16) entirely in favor of just
relying on the per-tensor type/size fields. This has no immediate
benefit, but makes it easier to experiment with different formats, and
should make it easier to support the new GPTQ-for-LLaMa models in the
future (I have some work in progress on that front).
- Support VirtualLock on Windows (using the same `--mlock` option as on
Unix).
- Indicate loading progress when using mmap + mlock. (Which led me
to the interesting observation that on my Linux machine, with a
warm file cache, mlock actually takes some time, whereas mmap
without mlock starts almost instantly...)
- To help implement this, move mlock support from ggml to the
loading code.
- madvise/PrefetchVirtualMemory support (based on #740)
- Switch from ifstream to the `fopen` family of functions to avoid
unnecessary copying and, when mmap is enabled, allow reusing the same
file descriptor for both metadata reads and mmap (whereas the existing
implementation opens the file a second time to mmap).
- Quantization now produces a single-file output even with multi-file
inputs (not really a feature as much as 'it was easier this way').
Implementation notes:
I tried to factor the code into more discrete pieces than before.
Regarding code style: I tried to follow the code style, but I'm naughty
and used a few advanced C++ features repeatedly:
- Destructors to make it easier to ensure everything gets cleaned up.
- Exceptions. I don't even usually use exceptions when writing C++, and
I can remove them if desired... but here they make the loading code
much more succinct while still properly handling a variety of errors,
ranging from API calls failing to integer overflow and allocation
failure. The exceptions are converted to error codes at the
API boundary.)
Co-authored-by: Pavol Rusnak <pavol@rusnak.io> (for the bit I copied from #740)
2023-04-08 21:24:37 +02:00
|
|
|
if (llama_mmap_supported()) {
|
2023-07-23 13:59:48 +02:00
|
|
|
fprintf(stdout, " --no-mmap do not memory-map model (slower load but may reduce pageouts if not using mlock)\n");
|
Rewrite loading code to try to satisfy everyone:
- Support all three formats (ggml, ggmf, ggjt). (However, I didn't
include the hack needed to support GPT4All files without conversion.
Those can still be used after converting them with convert.py from my
other PR.)
- Support both mmap and read (mmap is used by default, but can be
disabled with `--no-mmap`, and is automatically disabled for pre-ggjt
files or on platforms where mmap is not supported).
- Support multi-file models like before, but automatically determine the
number of parts rather than requiring `--n_parts`.
- Improve validation and error checking.
- Stop using the per-file type field (f16) entirely in favor of just
relying on the per-tensor type/size fields. This has no immediate
benefit, but makes it easier to experiment with different formats, and
should make it easier to support the new GPTQ-for-LLaMa models in the
future (I have some work in progress on that front).
- Support VirtualLock on Windows (using the same `--mlock` option as on
Unix).
- Indicate loading progress when using mmap + mlock. (Which led me
to the interesting observation that on my Linux machine, with a
warm file cache, mlock actually takes some time, whereas mmap
without mlock starts almost instantly...)
- To help implement this, move mlock support from ggml to the
loading code.
- madvise/PrefetchVirtualMemory support (based on #740)
- Switch from ifstream to the `fopen` family of functions to avoid
unnecessary copying and, when mmap is enabled, allow reusing the same
file descriptor for both metadata reads and mmap (whereas the existing
implementation opens the file a second time to mmap).
- Quantization now produces a single-file output even with multi-file
inputs (not really a feature as much as 'it was easier this way').
Implementation notes:
I tried to factor the code into more discrete pieces than before.
Regarding code style: I tried to follow the code style, but I'm naughty
and used a few advanced C++ features repeatedly:
- Destructors to make it easier to ensure everything gets cleaned up.
- Exceptions. I don't even usually use exceptions when writing C++, and
I can remove them if desired... but here they make the loading code
much more succinct while still properly handling a variety of errors,
ranging from API calls failing to integer overflow and allocation
failure. The exceptions are converted to error codes at the
API boundary.)
Co-authored-by: Pavol Rusnak <pavol@rusnak.io> (for the bit I copied from #740)
2023-04-08 21:24:37 +02:00
|
|
|
}
|
2023-07-23 13:59:48 +02:00
|
|
|
fprintf(stdout, " --numa attempt optimizations that help on some NUMA systems\n");
|
|
|
|
fprintf(stdout, " if run without this previously, it is recommended to drop the system page cache before using this\n");
|
|
|
|
fprintf(stdout, " see https://github.com/ggerganov/llama.cpp/issues/1437\n");
|
2023-05-28 19:48:57 +02:00
|
|
|
#ifdef LLAMA_SUPPORTS_GPU_OFFLOAD
|
2023-07-23 13:59:48 +02:00
|
|
|
fprintf(stdout, " -ngl N, --n-gpu-layers N\n");
|
|
|
|
fprintf(stdout, " number of layers to store in VRAM\n");
|
|
|
|
fprintf(stdout, " -ts SPLIT --tensor-split SPLIT\n");
|
|
|
|
fprintf(stdout, " how to split tensors across multiple GPUs, comma-separated list of proportions, e.g. 3,1\n");
|
2023-08-22 22:47:05 +02:00
|
|
|
fprintf(stdout, " -mg i, --main-gpu i the GPU to use for scratch and small tensors\n");
|
|
|
|
fprintf(stdout, " -lv, --low-vram don't allocate VRAM scratch buffer\n");
|
2023-08-25 11:09:42 +02:00
|
|
|
#ifdef GGML_USE_CUBLAS
|
2023-08-22 22:47:05 +02:00
|
|
|
fprintf(stdout, " -nommq, --no-mul-mat-q\n");
|
2023-08-25 11:09:42 +02:00
|
|
|
fprintf(stdout, " use " GGML_CUBLAS_NAME " instead of custom mul_mat_q " GGML_CUDA_NAME " kernels.\n");
|
2023-08-22 22:47:05 +02:00
|
|
|
fprintf(stdout, " Not recommended since this is both slower and uses more VRAM.\n");
|
2023-08-25 11:09:42 +02:00
|
|
|
#endif // GGML_USE_CUBLAS
|
2023-05-28 19:48:57 +02:00
|
|
|
#endif
|
2023-07-23 13:59:48 +02:00
|
|
|
fprintf(stdout, " --mtest compute maximum memory usage\n");
|
|
|
|
fprintf(stdout, " --export export the computation graph to 'llama.ggml'\n");
|
|
|
|
fprintf(stdout, " --verbose-prompt print prompt before generation\n");
|
2023-08-04 17:20:12 +02:00
|
|
|
fprintf(stderr, " --simple-io use basic IO for better compatibility in subprocesses and limited consoles\n");
|
2023-07-23 13:59:48 +02:00
|
|
|
fprintf(stdout, " --lora FNAME apply LoRA adapter (implies --no-mmap)\n");
|
|
|
|
fprintf(stdout, " --lora-base FNAME optional model to use as a base for the layers modified by the LoRA adapter\n");
|
|
|
|
fprintf(stdout, " -m FNAME, --model FNAME\n");
|
|
|
|
fprintf(stdout, " model path (default: %s)\n", params.model.c_str());
|
2023-08-28 17:59:39 +02:00
|
|
|
fprintf(stdout, " -ld LOGDIR, --logdir LOGDIR\n");
|
|
|
|
fprintf(stdout, " path under which to save YAML logs (no logging if unset)\n");
|
2023-07-23 13:59:48 +02:00
|
|
|
fprintf(stdout, "\n");
|
2023-03-10 19:40:58 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
std::string gpt_random_prompt(std::mt19937 & rng) {
|
|
|
|
const int r = rng() % 10;
|
|
|
|
switch (r) {
|
|
|
|
case 0: return "So";
|
|
|
|
case 1: return "Once upon a time";
|
|
|
|
case 2: return "When";
|
|
|
|
case 3: return "The";
|
|
|
|
case 4: return "After";
|
|
|
|
case 5: return "If";
|
|
|
|
case 6: return "import";
|
|
|
|
case 7: return "He";
|
|
|
|
case 8: return "She";
|
|
|
|
case 9: return "They";
|
|
|
|
default: return "To";
|
|
|
|
}
|
|
|
|
|
|
|
|
return "The";
|
|
|
|
}
|
|
|
|
|
2023-08-21 22:07:43 +02:00
|
|
|
//
|
|
|
|
// Model utils
|
|
|
|
//
|
2023-03-28 16:09:55 +02:00
|
|
|
|
2023-07-11 18:18:43 +02:00
|
|
|
struct llama_context_params llama_context_params_from_gpt_params(const gpt_params & params) {
|
2023-05-02 22:39:51 +02:00
|
|
|
auto lparams = llama_context_default_params();
|
|
|
|
|
2023-07-22 20:17:57 +02:00
|
|
|
lparams.n_ctx = params.n_ctx;
|
|
|
|
lparams.n_batch = params.n_batch;
|
|
|
|
lparams.n_gpu_layers = params.n_gpu_layers;
|
|
|
|
lparams.main_gpu = params.main_gpu;
|
|
|
|
lparams.tensor_split = params.tensor_split;
|
|
|
|
lparams.low_vram = params.low_vram;
|
2023-07-31 15:44:35 +02:00
|
|
|
lparams.mul_mat_q = params.mul_mat_q;
|
2023-07-22 20:17:57 +02:00
|
|
|
lparams.seed = params.seed;
|
|
|
|
lparams.f16_kv = params.memory_f16;
|
|
|
|
lparams.use_mmap = params.use_mmap;
|
|
|
|
lparams.use_mlock = params.use_mlock;
|
|
|
|
lparams.logits_all = params.perplexity;
|
|
|
|
lparams.embedding = params.embedding;
|
llama : add custom RoPE (#2054)
* Implement customizable RoPE
The original RoPE has pre-defined parameters
theta_i = 10000^(−2(i−1)/d), for i in [1, 2, ..., d/2]
Our customizable RoPE, ggml_rope_custom_inplace, uses
theta_i = scale * base^(−2(i−1)/d), for i in [1, 2, ..., d/2]
with the default matches the original
scale = 1.0
base = 10000
The new command line arguments
--rope-freq-base
--rope-freq-scale
set the two new RoPE parameter.
Recent researches show changing these two parameters extends the context limit with minimal loss.
1. Extending Context to 8K
kaiokendev
https://kaiokendev.github.io/til#extending-context-to-8k
2. Extending Context Window of Large Language Models via Positional Interpolation
Shouyuan Chen, Sherman Wong, Liangjian Chen, Yuandong Tian
https://arxiv.org/abs/2306.15595
3. NTK-Aware Scaled RoPE allows LLaMA models to have extended (8k+) context size without any fine-tuning and minimal perplexity degradation.
https://www.reddit.com/user/bloc97
https://www.reddit.com/r/LocalLLaMA/comments/14lz7j5/ntkaware_scaled_rope_allows_llama_models_to_have/
For the bold, try adding the following command line parameters to your favorite model:
-c 16384 --rope-freq-base 80000 --rope-freq-scale 0.5
* ggml-metal: fix custom rope
* common: fix argument names in help
* llama: increase MEM_REQ_EVAL for MODEL_3B
It avoids crashing for quantized weights on CPU.
Better ways to calculate the required buffer size would be better.
* llama: make MEM_REQ_EVAL depend on n_ctx
* server: use proper Content-Type in curl examples
Without the header Content-Type: application/json, curl will POST with
Content-Type: application/x-www-form-urlencoded
Though our simple server doesn't care, the httplib.h used has a limit
with CPPHTTPLIB_FORM_URL_ENCODED_PAYLOAD_MAX_LENGTH 8192
With Content-Type: application/json, we can send large json data.
* style : minor fixes, mostly indentations
* ggml : fix asserts
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-07-15 12:34:16 +02:00
|
|
|
lparams.rope_freq_base = params.rope_freq_base;
|
|
|
|
lparams.rope_freq_scale = params.rope_freq_scale;
|
2023-05-02 22:39:51 +02:00
|
|
|
|
2023-07-11 18:18:43 +02:00
|
|
|
return lparams;
|
|
|
|
}
|
|
|
|
|
2023-08-21 22:07:43 +02:00
|
|
|
std::tuple<struct llama_model *, struct llama_context *> llama_init_from_gpt_params(gpt_params & params) {
|
2023-07-11 18:18:43 +02:00
|
|
|
auto lparams = llama_context_params_from_gpt_params(params);
|
|
|
|
|
2023-06-24 10:47:58 +02:00
|
|
|
llama_model * model = llama_load_model_from_file(params.model.c_str(), lparams);
|
|
|
|
if (model == NULL) {
|
|
|
|
fprintf(stderr, "%s: error: failed to load model '%s'\n", __func__, params.model.c_str());
|
|
|
|
return std::make_tuple(nullptr, nullptr);
|
|
|
|
}
|
2023-05-02 22:39:51 +02:00
|
|
|
|
2023-06-24 10:47:58 +02:00
|
|
|
llama_context * lctx = llama_new_context_with_model(model, lparams);
|
2023-05-02 22:39:51 +02:00
|
|
|
if (lctx == NULL) {
|
2023-06-24 10:47:58 +02:00
|
|
|
fprintf(stderr, "%s: error: failed to create context with model '%s'\n", __func__, params.model.c_str());
|
|
|
|
llama_free_model(model);
|
|
|
|
return std::make_tuple(nullptr, nullptr);
|
2023-05-02 22:39:51 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
if (!params.lora_adapter.empty()) {
|
2023-06-24 10:47:58 +02:00
|
|
|
int err = llama_model_apply_lora_from_file(model,
|
2023-05-02 22:39:51 +02:00
|
|
|
params.lora_adapter.c_str(),
|
|
|
|
params.lora_base.empty() ? NULL : params.lora_base.c_str(),
|
|
|
|
params.n_threads);
|
|
|
|
if (err != 0) {
|
|
|
|
fprintf(stderr, "%s: error: failed to apply lora adapter\n", __func__);
|
2023-06-24 10:47:58 +02:00
|
|
|
llama_free(lctx);
|
|
|
|
llama_free_model(model);
|
|
|
|
return std::make_tuple(nullptr, nullptr);
|
2023-05-02 22:39:51 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-08-21 22:07:43 +02:00
|
|
|
if (params.ignore_eos) {
|
|
|
|
params.logit_bias[llama_token_eos(lctx)] = -INFINITY;
|
|
|
|
}
|
|
|
|
|
2023-06-24 10:47:58 +02:00
|
|
|
return std::make_tuple(model, lctx);
|
2023-05-02 22:39:51 +02:00
|
|
|
}
|
2023-08-21 22:07:43 +02:00
|
|
|
|
|
|
|
//
|
|
|
|
// Vocab utils
|
|
|
|
//
|
|
|
|
|
|
|
|
std::vector<llama_token> llama_tokenize(
|
|
|
|
struct llama_context * ctx,
|
|
|
|
const std::string & text,
|
|
|
|
bool add_bos) {
|
|
|
|
// upper limit for the number of tokens
|
|
|
|
int n_tokens = text.length() + add_bos;
|
|
|
|
std::vector<llama_token> result(n_tokens);
|
|
|
|
n_tokens = llama_tokenize(ctx, text.c_str(), result.data(), result.size(), add_bos);
|
|
|
|
if (n_tokens < 0) {
|
|
|
|
result.resize(-n_tokens);
|
|
|
|
int check = llama_tokenize(ctx, text.c_str(), result.data(), result.size(), add_bos);
|
|
|
|
GGML_ASSERT(check == -n_tokens);
|
|
|
|
} else {
|
|
|
|
result.resize(n_tokens);
|
|
|
|
}
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
2023-08-27 13:19:19 +02:00
|
|
|
std::string llama_token_to_piece(const struct llama_context * ctx, llama_token token) {
|
2023-08-21 22:07:43 +02:00
|
|
|
std::vector<char> result(8, 0);
|
2023-08-27 13:19:19 +02:00
|
|
|
const int n_tokens = llama_token_to_piece(ctx, token, result.data(), result.size());
|
2023-08-21 22:07:43 +02:00
|
|
|
if (n_tokens < 0) {
|
|
|
|
result.resize(-n_tokens);
|
2023-08-27 13:19:19 +02:00
|
|
|
int check = llama_token_to_piece(ctx, token, result.data(), result.size());
|
2023-08-21 22:07:43 +02:00
|
|
|
GGML_ASSERT(check == -n_tokens);
|
|
|
|
} else {
|
|
|
|
result.resize(n_tokens);
|
|
|
|
}
|
|
|
|
|
|
|
|
return std::string(result.data(), result.size());
|
|
|
|
}
|
2023-08-27 13:19:19 +02:00
|
|
|
|
|
|
|
std::string llama_detokenize_spm(llama_context * ctx, const std::vector<llama_token> & tokens) {
|
|
|
|
const llama_token bos_id = llama_token_bos(ctx);
|
|
|
|
|
|
|
|
std::string piece;
|
|
|
|
std::string result;
|
|
|
|
|
|
|
|
for (size_t i = 0; i < tokens.size(); ++i) {
|
|
|
|
piece = llama_token_to_piece(ctx, tokens[i]);
|
|
|
|
|
|
|
|
// remove the leading space of the first non-BOS token
|
|
|
|
if (((tokens[0] == bos_id && i == 1) || (tokens[0] != bos_id && i == 0)) && piece[0] == ' ') {
|
|
|
|
piece = piece.substr(1);
|
|
|
|
}
|
|
|
|
|
|
|
|
result += piece;
|
|
|
|
}
|
|
|
|
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
|
|
|
std::string llama_detokenize_bpe(llama_context * ctx, const std::vector<llama_token> & tokens) {
|
|
|
|
std::string piece;
|
|
|
|
std::string result;
|
|
|
|
|
|
|
|
for (size_t i = 0; i < tokens.size(); ++i) {
|
|
|
|
piece = llama_token_to_piece(ctx, tokens[i]);
|
|
|
|
|
|
|
|
result += piece;
|
|
|
|
}
|
|
|
|
|
|
|
|
return result;
|
|
|
|
}
|
2023-08-28 17:59:39 +02:00
|
|
|
|
|
|
|
// returns true if successful, false otherwise
|
|
|
|
bool create_directory_with_parents(const std::string & path) {
|
|
|
|
#ifdef _WIN32
|
|
|
|
std::wstring_convert<std::codecvt_utf8<wchar_t>> converter;
|
|
|
|
std::wstring wpath = converter.from_bytes(path);
|
|
|
|
|
|
|
|
// if the path already exists, check whether it's a directory
|
|
|
|
const DWORD attributes = GetFileAttributesW(wpath.c_str());
|
|
|
|
if ((attributes != INVALID_FILE_ATTRIBUTES) && (attributes & FILE_ATTRIBUTE_DIRECTORY)) {
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
size_t pos_slash = 0;
|
|
|
|
|
|
|
|
// process path from front to back, procedurally creating directories
|
|
|
|
while ((pos_slash = path.find('\\', pos_slash)) != std::string::npos) {
|
|
|
|
const std::wstring subpath = wpath.substr(0, pos_slash);
|
|
|
|
const wchar_t * test = subpath.c_str();
|
|
|
|
|
|
|
|
const bool success = CreateDirectoryW(test, NULL);
|
|
|
|
if (!success) {
|
|
|
|
const DWORD error = GetLastError();
|
|
|
|
|
|
|
|
// if the path already exists, ensure that it's a directory
|
|
|
|
if (error == ERROR_ALREADY_EXISTS) {
|
|
|
|
const DWORD attributes = GetFileAttributesW(subpath.c_str());
|
|
|
|
if (attributes == INVALID_FILE_ATTRIBUTES || !(attributes & FILE_ATTRIBUTE_DIRECTORY)) {
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
pos_slash += 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
return true;
|
|
|
|
#else
|
|
|
|
// if the path already exists, check whether it's a directory
|
|
|
|
struct stat info;
|
|
|
|
if (stat(path.c_str(), &info) == 0) {
|
|
|
|
return S_ISDIR(info.st_mode);
|
|
|
|
}
|
|
|
|
|
|
|
|
size_t pos_slash = 1; // skip leading slashes for directory creation
|
|
|
|
|
|
|
|
// process path from front to back, procedurally creating directories
|
|
|
|
while ((pos_slash = path.find('/', pos_slash)) != std::string::npos) {
|
|
|
|
const std::string subpath = path.substr(0, pos_slash);
|
|
|
|
struct stat info;
|
|
|
|
|
|
|
|
// if the path already exists, ensure that it's a directory
|
|
|
|
if (stat(subpath.c_str(), &info) == 0) {
|
|
|
|
if (!S_ISDIR(info.st_mode)) {
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
// create parent directories
|
|
|
|
const int ret = mkdir(subpath.c_str(), 0755);
|
|
|
|
if (ret != 0) {
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
pos_slash += 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
return true;
|
|
|
|
#endif // _WIN32
|
|
|
|
}
|
|
|
|
|
|
|
|
void dump_vector_float_yaml(FILE * stream, const char * prop_name, const std::vector<float> & data) {
|
|
|
|
if (data.empty()) {
|
|
|
|
fprintf(stream, "%s:\n", prop_name);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
fprintf(stream, "%s: [", prop_name);
|
|
|
|
for (size_t i = 0; i < data.size() - 1; ++i) {
|
|
|
|
fprintf(stream, "%e, ", data[i]);
|
|
|
|
}
|
|
|
|
fprintf(stream, "%e]\n", data.back());
|
|
|
|
}
|
|
|
|
|
|
|
|
void dump_vector_int_yaml(FILE * stream, const char * prop_name, const std::vector<int> & data) {
|
|
|
|
if (data.empty()) {
|
|
|
|
fprintf(stream, "%s:\n", prop_name);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
fprintf(stream, "%s: [", prop_name);
|
|
|
|
for (size_t i = 0; i < data.size() - 1; ++i) {
|
|
|
|
fprintf(stream, "%d, ", data[i]);
|
|
|
|
}
|
|
|
|
fprintf(stream, "%d]\n", data.back());
|
|
|
|
}
|
|
|
|
|
|
|
|
void dump_string_yaml_multiline(FILE * stream, const char * prop_name, const char * data) {
|
|
|
|
std::string data_str(data == NULL ? "" : data);
|
|
|
|
|
|
|
|
if (data_str.empty()) {
|
|
|
|
fprintf(stream, "%s:\n", prop_name);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
size_t pos_start = 0;
|
|
|
|
size_t pos_found = 0;
|
|
|
|
|
|
|
|
if (!data_str.empty() && (std::isspace(data_str[0]) || std::isspace(data_str.back()))) {
|
|
|
|
data_str = std::regex_replace(data_str, std::regex("\n"), "\\n");
|
|
|
|
data_str = std::regex_replace(data_str, std::regex("\""), "\\\"");
|
|
|
|
data_str = "\"" + data_str + "\"";
|
|
|
|
fprintf(stream, "%s: %s\n", prop_name, data_str.c_str());
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (data_str.find('\n') == std::string::npos) {
|
|
|
|
fprintf(stream, "%s: %s\n", prop_name, data_str.c_str());
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
fprintf(stream, "%s: |\n", prop_name);
|
|
|
|
while ((pos_found = data_str.find('\n', pos_start)) != std::string::npos) {
|
|
|
|
fprintf(stream, " %s\n", data_str.substr(pos_start, pos_found-pos_start).c_str());
|
|
|
|
pos_start = pos_found + 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
std::string get_sortable_timestamp() {
|
|
|
|
using clock = std::chrono::system_clock;
|
|
|
|
|
|
|
|
const clock::time_point current_time = clock::now();
|
|
|
|
const time_t as_time_t = clock::to_time_t(current_time);
|
|
|
|
char timestamp_no_ns[100];
|
|
|
|
std::strftime(timestamp_no_ns, 100, "%Y_%m_%d-%H_%M_%S", std::localtime(&as_time_t));
|
|
|
|
|
|
|
|
const int64_t ns = std::chrono::duration_cast<std::chrono::nanoseconds>(
|
|
|
|
current_time.time_since_epoch() % 1000000000).count();
|
train : mem usage and other improvements (#2439)
* fix track_max_mem in forward_batch_wo_cache_flash_attn_train
* remove unnecessary Adam(W) optimizer tensors.
reduces optimizer memory overhead from 7*modelsize to 2*modelsize.
additionally allows to optimize models with more than 2^31 parameters by replacing int with int64_t.
bumps training checkpoint file version, but old checkpoints can still be read.
new version with less tensors is saved.
* add gradient clipping to AdamW
* Fix reset of unused g->nodes and g->grads to NULL
* implement gradient checkpointing for training
reduces memory overhead from O(n_layer) to O(sqrt(n_layer))
as explained in readme of https://github.com/cybertronai/gradient-checkpointing
* remove unused compute buffer 3
* add and use function ggml_build_backward_expand to avoid stack overflows with large maximum number of nodes
GGML_API void ggml_build_backward_expand(struct ggml_context * ctx, struct ggml_cgraph * gf, struct ggml_cgraph * gb, bool keep);
* change AdamW decay parameter to work like the torch AdamW decay parameter
It is now relative to Adam learning rate `alpha*sched`.
Before that it was relative to `sched` only.
`alpha` being the maximum learning rate and `sched` being a scaling parameter in [0..1]
* change default AdamW weight decay parameter used in training to 0.1 as used in nanoGPT
* change default AdamW weight decay parameter defined in ggml to 0.0, making Adam default instead of AdamW
btw: the default weight decay parameter for torch.optim.AdamW is 0.01
* bug fixes for cross entropy loss
ggml_cross_entropy_loss: sums where not correctly added in workload of each thread
ggml_cross_entropy_loss_back: simplify backward process, reducing numerical issues
guard usage of exp f16 lookup in cross entropy by #define GGML_CROSS_ENTROPY_EXP_FP16
cross entropy loss is only used once during training, but it is quite sensitive to numerical errors introduced by exp-f16-lookup.
so exp-f16-lookup for cross entropy loss is disabled by default, trading better gradients for very slightly worse runtime performance.
* fix test-grad0 for cross_entropy_loss
the second argument to cross_entropy_loss must sum up to 1 for each row
* fix test-grad0 for soft_max
dont use only sum as aggregation, because sum of softmax is always 1 -> finite differences should not work
instead use sum(log(soft_max()*(1-eps)+eps)); use eps to avoid log(0)
* improve finite differences of test-grad0 by using double instead of float
* change cross_entropy_loss to output average over all rows
this helps keeping the loss and gradients in a sane range
* improve gradient checkpointing
sqrt(n_layers) is only the best checkpoint step when mem size of checkpoints and mem size of layers are equal.
since layers require more memory than the single-tensor-checkpoint we use, the optimal values are compute different:
```
given: n, u, v
objective: minimize(a*u+b*v) where a*b=n, a>0, b>0
b=n/a
minimize(a*u+v*n/a)
diff(a*u+v*n/a, a) = u - (v*n/a)/a
diff(a*u+v*n/a, a) == 0
u - (v*n/a)/a == 0
u == v*n/(a*a)
u*a*a = v*n
a*a = v*n/u
a = sqrt(n*v/u)
```
this change results in more checkpoints, requiring less layers to store between checkpoints, overall improving memory usage.
* disable gradient checkpointing debug output
* llama : fix rope usage in train-text-from-scratch after ChatGLM change
* add more training parameters:
--enable-restart N Only for Adam optimizer. Enable restarts of cos-decay
--disable-restart N Only for Adam optimizer. Disable restarts of cos-decay
--opt-past N Number of optimization iterations to track for delta convergence test. Disabled when zero.
--opt-delta N Maximum delta for delta convergence test. Disabled when <= zero.
--opt-max-no-improvement N Maximum number of optimization iterations with no improvement. Disabled when <= zero.
--adam-epsf N AdamW epsilon for convergence test. Disabled when <= zero.
--adam-min-alpha N Adam minimum learning rate alpha, usually 0.1 * alpha
* replace memcpy with reshape operation so that the graph is not cut at the input
this makes it possible to store other values into the input tensor and then simply recompute the graph without rebuilding it
* remove unused function argument from get_example_targets_batch
* measure and print total training time
* add optimization callback to ggml_opt_resume_g
this callback is called before each iteration with custom data and pointer to learning schedule parameter (only used in Adam(W)).
can be used for dynamic learning schedule and setting input data for batches before each iteration
* use optimization callback in training
allows dynamic learning schedule and different batch data for each iteration without relying on low n_iter and high n_examples parameters
reduces runtime by avoiding restart of optimization function and improves training convergence by providing a different batch for each iteration
* add minimum number of tensor dimensions to apply weight decay (default 2)
this allows to not apply weight decay to bias parameters
* rename training parameter cos-decay-alpha to cos-decay-min and clarify that adam-min-alpha also applies to warmup
* fix increase of model.train_samples and model.train_tokens
now that each optimizer iteration gets its own batch we need to multiply by number of opt iterations
* change sampling parameters for prediction after training to defaults of common.h
and clarify what is context for prediction and what are generated tokens
* tighten abs error bounds for cross_entropy_loss in test-grad0
* add conditional compilation of using F16 exp in flash attention
uncomment `// #define GGML_FLASH_ATTN_EXP_FP16` to enable usage of f16 exp in flash attention
* tighten abs error bounds for flash_attn in test-grad0
* tighten abs error bounds for sqrt in test-grad0
* remove out-commented vectorized code of opt_adam
the vectorized code might be bit faster for low number of parameters, but it had a big memory usage overhead
* ggml : update ggml_rms_norm_back with configurable eps
* llama training : fix ggml_rms_norm_back calls to pass configurable eps
* remove trailing whitespace
* add train function using automatic gradient checkpointing backward pass and allocator
* in train function replace add_inplace by regular add
because using add_inplace seems to result in different gradients
* don't use allocate hash_map on context
because the context has no_alloc=True when using memory allocator resulting in NULL data pointers
* correctly clone reshape and permute operations by also cloning tensor->nb values
* fix variable name and add missing type cast
* terminate recursive tensor cloning when reaching tensor without src tensors
* correctly clone view tensors by setting data pointers
without this the checkpointing would only work when being used together with memory allocator
* fix variable names
* swap arguments to commutative ops to be the same as in `forward_batch_wo_cache_flash_attn`
* add input tensors as checkpoints
so that recursive tensor cloning of gradient checkpointing terminates on input tensors
* fix variable name and add missing boolean negation
* make sure some tensors are not reallocated by inserting new temporary nodes depending on them:
output and parameter gradient tensors need to be available at the end of the graph execution
parameter gradient tensors also need to be available before the graph execution because they are set to zero before each optimizer iteration
checkpoint tensors are allocated all together to reduce memory allocator fragmentation
afterwards, in addition to the temporary nodes, we also need to reset the temporary leafs
* fix ASSERT to work with zero layers
* add training options whether to use allocator and/or unified training function
* integrate unified training function which may use memory allocator
the unified training function also supports arguments whether to use flash attention and/or gradient checkpointing
* format name of cloned tensors with " (clone)" suffix
* set names for tensors in unified train function for easier debugging
* allocate graph on context using ggml_new_graph
* remove handwritten training functions
* remove unused training parameters "use_scratch" and "use_unified"
* remove trailing whitespace
* remove unused train params: mem_compute1_gb & mem_compute2_gb
mem_compute_gb is used for compute when automatic memory allocator is not enabled, otherwise it can be very small to only hold the tensor definitions
mem_compute0_gb is used for automatic memory allocator (as long as measurement of max required size is not implemented)
* remove unused forward_batch function
* add debug asserts in ggml_allocr_alloc to some common pitfalls when using this function directly
* only use ggml_allocr_alloc when tensor has NULL data and is no view
* fix test when to create temporary backward graph
temporary backward graph is only necessary when using checkpointing
* fix memory "leak" in optimizers
each iteration a new cplan with new memory for work data was allocated.
now cplan creation only happens at the start of optimization, with each iteration reusing the cplan and its work data.
* reverse order of for loop in ggml_build_backward_expand to save memory when using gradient checkpointing and allocator
with this loop order gradient checkpointing with allocator on 16 layer model saves 13% memory; 2 layer memory it saves 2% memory.
the computation results are the same
* add missing lctx argument to get_example_targets_batch
* implement llama model file saving using gguf
checkpoint loading and saving disabled, to be replaced by loading and saving via gguf
* implement loading/saving of checkpointing files using GGUF
* bug fixes
* add checkpoint file version for future compatibility
* update readme with gguf filenames
* save & load opt->just_initialized value
* add first draft for checkpoint conversion script
* add gguf arch and ftype
* save opt parameter counter as uint64
* add gguf key and tensor names for optimizer and training
* add layer_norm_rms_eps to checkpoint convert script
* use same GGUF_GET_KEY macro as in llama.cpp
* use norm_rms_eps, and rope parameters and command line options to set them
* fix memory corruption bug in gguf
ctx->kv and ctx->infos was reallocated using not-aligned realloc, but freed with aligned free.
to fix this a GGML_ALIGNED_REALLOC was added, but there is no posix_memalign_realloc function.
so on non-windows and non-mingw32 platforms we fall back to aligned malloc, followed by copying
and freeing the old data.
* add gguf example cmake file
* bug fixes in tokenize_file
* bug fixes in load_llama_model_gguf
* bug fix: init model when no checkpoint was loaded
* bug fix in read_tensor_by_name
* bug fix in load_opt_context_gguf
* avoid printing lots of spaced on the unusual case that loss gets nan
* set name of tensors with empty name from what was read from gguf
* remove trailing whitespace
* print data checksums before saving and after loading to verify correctness
* bug fixes for convert-train-checkpoint-to-gguf
* temporarily add code to write old checkpoint files
used to verify that old checkpoint files are correctly converted to gguf
* bug fixes for convert-train-checkpoint-to-gguf.py loading checkpoints with opt_version=0
* remove code used to verify correctness of checkpoint file conversion
* remove trailing whitespace
* remove prediction related code
use main for prediction, it is better optimized
* update train-text-from-scratch README.md
* fix non-windows GGML_ALIGNED_REALLOC
* add missing blank line at end of file
* remove GGML_ALIGNED_REALLOC and use normal malloc/realloc/free for gguf ctx->kv & ctx->infos
* train : fix compile warnings
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-08-28 21:51:47 +02:00
|
|
|
char timestamp_ns[11];
|
|
|
|
snprintf(timestamp_ns, 11, "%09" PRId64, ns);
|
2023-08-28 17:59:39 +02:00
|
|
|
|
|
|
|
return std::string(timestamp_no_ns) + "." + std::string(timestamp_ns);
|
|
|
|
}
|
|
|
|
|
|
|
|
void dump_non_result_info_yaml(FILE * stream, const gpt_params & params, const llama_context * lctx,
|
|
|
|
const std::string & timestamp, const std::vector<int> & prompt_tokens, const char * model_desc) {
|
|
|
|
fprintf(stream, "build_commit: %s\n", BUILD_COMMIT);
|
|
|
|
fprintf(stream, "build_number: %d\n", BUILD_NUMBER);
|
|
|
|
fprintf(stream, "cpu_has_arm_fma: %s\n", ggml_cpu_has_arm_fma() ? "true" : "false");
|
|
|
|
fprintf(stream, "cpu_has_avx: %s\n", ggml_cpu_has_avx() ? "true" : "false");
|
|
|
|
fprintf(stream, "cpu_has_avx2: %s\n", ggml_cpu_has_avx2() ? "true" : "false");
|
|
|
|
fprintf(stream, "cpu_has_avx512: %s\n", ggml_cpu_has_avx512() ? "true" : "false");
|
|
|
|
fprintf(stream, "cpu_has_avx512_vbmi: %s\n", ggml_cpu_has_avx512_vbmi() ? "true" : "false");
|
|
|
|
fprintf(stream, "cpu_has_avx512_vnni: %s\n", ggml_cpu_has_avx512_vnni() ? "true" : "false");
|
|
|
|
fprintf(stream, "cpu_has_blas: %s\n", ggml_cpu_has_blas() ? "true" : "false");
|
|
|
|
fprintf(stream, "cpu_has_cublas: %s\n", ggml_cpu_has_cublas() ? "true" : "false");
|
|
|
|
fprintf(stream, "cpu_has_clblast: %s\n", ggml_cpu_has_clblast() ? "true" : "false");
|
|
|
|
fprintf(stream, "cpu_has_fma: %s\n", ggml_cpu_has_fma() ? "true" : "false");
|
|
|
|
fprintf(stream, "cpu_has_gpublas: %s\n", ggml_cpu_has_gpublas() ? "true" : "false");
|
|
|
|
fprintf(stream, "cpu_has_neon: %s\n", ggml_cpu_has_neon() ? "true" : "false");
|
|
|
|
fprintf(stream, "cpu_has_f16c: %s\n", ggml_cpu_has_f16c() ? "true" : "false");
|
|
|
|
fprintf(stream, "cpu_has_fp16_va: %s\n", ggml_cpu_has_fp16_va() ? "true" : "false");
|
|
|
|
fprintf(stream, "cpu_has_wasm_simd: %s\n", ggml_cpu_has_wasm_simd() ? "true" : "false");
|
|
|
|
fprintf(stream, "cpu_has_blas: %s\n", ggml_cpu_has_blas() ? "true" : "false");
|
|
|
|
fprintf(stream, "cpu_has_sse3: %s\n", ggml_cpu_has_sse3() ? "true" : "false");
|
|
|
|
fprintf(stream, "cpu_has_vsx: %s\n", ggml_cpu_has_vsx() ? "true" : "false");
|
|
|
|
|
|
|
|
#ifdef NDEBUG
|
|
|
|
fprintf(stream, "debug: false\n");
|
|
|
|
#else
|
|
|
|
fprintf(stream, "debug: true\n");
|
|
|
|
#endif // NDEBUG
|
|
|
|
|
|
|
|
fprintf(stream, "model_desc: %s\n", model_desc);
|
|
|
|
fprintf(stream, "n_vocab: %d # output size of the final layer, 32001 for some models\n", llama_n_vocab(lctx));
|
|
|
|
|
|
|
|
#ifdef __OPTIMIZE__
|
|
|
|
fprintf(stream, "optimize: true\n");
|
|
|
|
#else
|
|
|
|
fprintf(stream, "optimize: false\n");
|
|
|
|
#endif // __OPTIMIZE__
|
|
|
|
|
|
|
|
fprintf(stream, "time: %s\n", timestamp.c_str());
|
|
|
|
|
|
|
|
fprintf(stream, "\n");
|
|
|
|
fprintf(stream, "###############\n");
|
|
|
|
fprintf(stream, "# User Inputs #\n");
|
|
|
|
fprintf(stream, "###############\n");
|
|
|
|
fprintf(stream, "\n");
|
|
|
|
|
|
|
|
fprintf(stream, "alias: %s # default: unknown\n", params.model_alias.c_str());
|
|
|
|
fprintf(stream, "batch_size: %d # default: 512\n", params.n_batch);
|
|
|
|
dump_string_yaml_multiline(stream, "cfg_negative_prompt", params.cfg_negative_prompt.c_str());
|
|
|
|
fprintf(stream, "cfg_scale: %f # default: 1.0\n", params.cfg_scale);
|
|
|
|
fprintf(stream, "chunks: %d # default: -1 (unlimited)\n", params.n_chunks);
|
|
|
|
fprintf(stream, "color: %s # default: false\n", params.use_color ? "true" : "false");
|
|
|
|
fprintf(stream, "ctx_size: %d # default: 512\n", params.n_ctx);
|
|
|
|
fprintf(stream, "escape: %s # default: false\n", params.escape ? "true" : "false");
|
|
|
|
fprintf(stream, "export: %s # default: false\n", params.export_cgraph ? "true" : "false");
|
|
|
|
fprintf(stream, "file: # never logged, see prompt instead. Can still be specified for input.\n");
|
|
|
|
fprintf(stream, "frequency_penalty: %f # default: 0.0 \n", params.frequency_penalty);
|
|
|
|
dump_string_yaml_multiline(stream, "grammar", params.grammar.c_str());
|
|
|
|
fprintf(stream, "grammar-file: # never logged, see grammar instead. Can still be specified for input.\n");
|
|
|
|
fprintf(stream, "hellaswag: %s # default: false\n", params.hellaswag ? "true" : "false");
|
|
|
|
fprintf(stream, "hellaswag_tasks: %ld # default: 400\n", params.hellaswag_tasks);
|
|
|
|
|
|
|
|
const auto logit_bias_eos = params.logit_bias.find(llama_token_eos(lctx));
|
|
|
|
const bool ignore_eos = logit_bias_eos != params.logit_bias.end() && logit_bias_eos->second == -INFINITY;
|
|
|
|
fprintf(stream, "ignore_eos: %s # default: false\n", ignore_eos ? "true" : "false");
|
|
|
|
|
|
|
|
dump_string_yaml_multiline(stream, "in_prefix", params.input_prefix.c_str());
|
|
|
|
fprintf(stream, "in_prefix_bos: %s # default: false\n", params.input_prefix_bos ? "true" : "false");
|
|
|
|
dump_string_yaml_multiline(stream, "in_suffix", params.input_prefix.c_str());
|
|
|
|
fprintf(stream, "instruct: %s # default: false\n", params.instruct ? "true" : "false");
|
|
|
|
fprintf(stream, "interactive: %s # default: false\n", params.interactive ? "true" : "false");
|
|
|
|
fprintf(stream, "interactive_first: %s # default: false\n", params.interactive_first ? "true" : "false");
|
|
|
|
fprintf(stream, "keep: %d # default: 0\n", params.n_keep);
|
|
|
|
fprintf(stream, "logdir: %s # default: unset (no logging)\n", params.logdir.c_str());
|
|
|
|
|
|
|
|
fprintf(stream, "logit_bias:\n");
|
|
|
|
for (std::pair<llama_token, float> lb : params.logit_bias) {
|
|
|
|
if (ignore_eos && lb.first == logit_bias_eos->first) {
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
fprintf(stream, " %d: %f", lb.first, lb.second);
|
|
|
|
}
|
|
|
|
|
|
|
|
fprintf(stream, "lora: %s\n", params.lora_adapter.c_str());
|
|
|
|
fprintf(stream, "lora_base: %s\n", params.lora_base.c_str());
|
|
|
|
fprintf(stream, "low_vram: %s # default: false\n", params.low_vram ? "true" : "false");
|
|
|
|
fprintf(stream, "main_gpu: %d # default: 0\n", params.main_gpu);
|
|
|
|
fprintf(stream, "memory_f32: %s # default: false\n", !params.memory_f16 ? "true" : "false");
|
|
|
|
fprintf(stream, "mirostat: %d # default: 0 (disabled)\n", params.mirostat);
|
|
|
|
fprintf(stream, "mirostat_ent: %f # default: 5.0\n", params.mirostat_tau);
|
|
|
|
fprintf(stream, "mirostat_lr: %f # default: 0.1\n", params.mirostat_eta);
|
|
|
|
fprintf(stream, "mlock: %s # default: false\n", params.use_mlock ? "true" : "false");
|
|
|
|
fprintf(stream, "model: %s # default: models/7B/ggml-model.bin\n", params.model.c_str());
|
|
|
|
fprintf(stream, "mtest: %s # default: false\n", params.mem_test ? "true" : "false");
|
|
|
|
fprintf(stream, "multiline_input: %s # default: false\n", params.multiline_input ? "true" : "false");
|
|
|
|
fprintf(stream, "n_gpu_layers: %d # default: 0\n", params.n_gpu_layers);
|
|
|
|
fprintf(stream, "n_predict: %d # default: -1 (unlimited)\n", params.n_predict);
|
|
|
|
fprintf(stream, "n_probs: %d # only used by server binary, default: 0\n", params.n_probs);
|
|
|
|
fprintf(stream, "no_mmap: %s # default: false\n", !params.use_mmap ? "true" : "false");
|
|
|
|
fprintf(stream, "no_mul_mat_q: %s # default: false\n", !params.mul_mat_q ? "true" : "false");
|
|
|
|
fprintf(stream, "no_penalize_nl: %s # default: false\n", !params.penalize_nl ? "true" : "false");
|
|
|
|
fprintf(stream, "numa: %s # default: false\n", params.numa ? "true" : "false");
|
|
|
|
fprintf(stream, "ppl_output_type: %d # default: 0\n", params.ppl_output_type);
|
|
|
|
fprintf(stream, "ppl_stride: %d # default: 0\n", params.ppl_stride);
|
|
|
|
fprintf(stream, "presence_penalty: %f # default: 0.0\n", params.presence_penalty);
|
|
|
|
dump_string_yaml_multiline(stream, "prompt", params.prompt.c_str());
|
|
|
|
fprintf(stream, "prompt_cache: %s\n", params.path_prompt_cache.c_str());
|
|
|
|
fprintf(stream, "prompt_cache_all: %s # default: false\n", params.prompt_cache_all ? "true" : "false");
|
|
|
|
fprintf(stream, "prompt_cache_ro: %s # default: false\n", params.prompt_cache_ro ? "true" : "false");
|
|
|
|
dump_vector_int_yaml(stream, "prompt_tokens", prompt_tokens);
|
|
|
|
fprintf(stream, "random_prompt: %s # default: false\n", params.random_prompt ? "true" : "false");
|
|
|
|
fprintf(stream, "repeat_penalty: %f # default: 1.1\n", params.repeat_penalty);
|
|
|
|
|
|
|
|
fprintf(stream, "reverse_prompt:\n");
|
|
|
|
for (std::string ap : params.antiprompt) {
|
|
|
|
size_t pos = 0;
|
|
|
|
while ((pos = ap.find('\n', pos)) != std::string::npos) {
|
|
|
|
ap.replace(pos, 1, "\\n");
|
|
|
|
pos += 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
fprintf(stream, " - %s\n", ap.c_str());
|
|
|
|
}
|
|
|
|
|
|
|
|
fprintf(stream, "rope_freq_base: %f # default: 10000.0\n", params.rope_freq_base);
|
|
|
|
fprintf(stream, "rope_freq_scale: %f # default: 1.0\n", params.rope_freq_scale);
|
|
|
|
fprintf(stream, "seed: %d # default: -1 (random seed)\n", params.seed);
|
|
|
|
fprintf(stream, "simple_io: %s # default: false\n", params.simple_io ? "true" : "false");
|
|
|
|
fprintf(stream, "temp: %f # default: 0.8\n", params.temp);
|
|
|
|
|
|
|
|
const std::vector<float> tensor_split_vector(params.tensor_split, params.tensor_split + LLAMA_MAX_DEVICES);
|
|
|
|
dump_vector_float_yaml(stream, "tensor_split", tensor_split_vector);
|
|
|
|
|
|
|
|
fprintf(stream, "tfs: %f # default: 1.0\n", params.tfs_z);
|
|
|
|
fprintf(stream, "threads: %d # default: %d\n", params.n_threads, std::thread::hardware_concurrency());
|
|
|
|
fprintf(stream, "top_k: %d # default: 40\n", params.top_k);
|
|
|
|
fprintf(stream, "top_p: %f # default: 0.95\n", params.top_p);
|
|
|
|
fprintf(stream, "typical_p: %f # default: 1.0\n", params.typical_p);
|
|
|
|
fprintf(stream, "verbose_prompt: %s # default: false\n", params.verbose_prompt ? "true" : "false");
|
|
|
|
}
|