Mirrors/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-10-29 22:20:15 +01:00
Code Issues Packages Projects Releases Wiki Activity

llama : rename batch_all to batch (#8881)

Browse Source 
This commit addresses the TODO in the code to rename the `batch_all`
parameter to `batch` in `llama_decode_internal`.
This commit is contained in:
Daniel Bevenius 2024-10-18 01:41:51 +02:00 committed by GitHub
parent 17bb928080
commit 6f55bccbb8
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 9 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
src/llama.cpp
View File

@ -17134,10 +17134,10 @@ static void llama_graph_compute(
// //
static int llama_decode_internal( static int llama_decode_internal(
llama_context & lctx, llama_context & lctx,
llama_batch batch_all) { // TODO: rename back to batch llama_batch batch) {
lctx.is_encoding = false; lctx.is_encoding = false;
const uint32_t n_tokens_all = batch_all.n_tokens; const uint32_t n_tokens_all = batch.n_tokens;
if (n_tokens_all == 0) { if (n_tokens_all == 0) {
LLAMA_LOG_ERROR("%s: n_tokens == 0\n", __func__); LLAMA_LOG_ERROR("%s: n_tokens == 0\n", __func__);
@ -17148,12 +17148,12 @@ static int llama_decode_internal(
const auto & hparams = model.hparams; const auto & hparams = model.hparams;
const auto & cparams = lctx.cparams; const auto & cparams = lctx.cparams;
GGML_ASSERT((!batch_all.token && batch_all.embd) || (batch_all.token && !batch_all.embd)); // NOLINT GGML_ASSERT((!batch.token && batch.embd) || (batch.token && !batch.embd)); // NOLINT
if (batch_all.token) { if (batch.token) {
for (uint32_t i = 0; i < n_tokens_all; ++i) { for (uint32_t i = 0; i < n_tokens_all; ++i) {
if (batch_all.token[i] < 0 || (uint32_t)batch_all.token[i] >= model.vocab.n_vocab) { if (batch.token[i] < 0 || (uint32_t)batch.token[i] >= model.vocab.n_vocab) {
LLAMA_LOG_ERROR("%s: invalid token[%d] = %d\n", __func__, i, batch_all.token[i]); LLAMA_LOG_ERROR("%s: invalid token[%d] = %d\n", __func__, i, batch.token[i]);
return -1; return -1;
} }
} }
@ -17184,9 +17184,9 @@ static int llama_decode_internal(
lctx.embd_seq.clear(); lctx.embd_seq.clear();
// count outputs // count outputs
if (batch_all.logits && !embd_pooled) { if (batch.logits && !embd_pooled) {
for (uint32_t i = 0; i < n_tokens_all; ++i) { for (uint32_t i = 0; i < n_tokens_all; ++i) {
n_outputs += batch_all.logits[i] != 0; n_outputs += batch.logits[i] != 0;
} }
} else if (lctx.logits_all || embd_pooled) { } else if (lctx.logits_all || embd_pooled) {
n_outputs = n_tokens_all; n_outputs = n_tokens_all;
@ -17195,7 +17195,7 @@ static int llama_decode_internal(
n_outputs = 1; n_outputs = 1;
} }
lctx.sbatch.from_batch(batch_all, n_embd, lctx.sbatch.from_batch(batch, n_embd,
/* simple_split */ !kv_self.recurrent, /* simple_split */ !kv_self.recurrent,
/* logits_all */ n_outputs == n_tokens_all); /* logits_all */ n_outputs == n_tokens_all);