2024-07-05 18:08:32 +02:00
# Add a new model architecture to `llama.cpp`
2024-04-10 08:58:48 +02:00
Adding a model requires few steps:
1. Convert the model to GGUF
2. Define the model architecture in `llama.cpp`
3. Build the GGML graph implementation
After following these steps, you can open PR.
Also, it is important to check that the examples and main ggml backends (CUDA, METAL, CPU) are working with the new architecture, especially:
2024-07-15 13:46:39 +02:00
- [main ](/examples/main/ )
- [imatrix ](/examples/imatrix/ )
- [quantize ](/examples/quantize/ )
- [server ](/examples/server/ )
2024-04-10 08:58:48 +02:00
### 1. Convert the model to GGUF
This step is done in python with a `convert` script using the [gguf ](https://pypi.org/project/gguf/ ) library.
2024-07-15 13:46:39 +02:00
Depending on the model architecture, you can use either [convert_hf_to_gguf.py ](/convert_hf_to_gguf.py ) or [examples/convert_legacy_llama.py ](/examples/convert_legacy_llama.py ) (for `llama/llama2` models in `.pth` format).
2024-04-10 08:58:48 +02:00
The convert script reads the model configuration, tokenizer, tensor names+data and converts them to GGUF metadata and tensors.
The required steps to implement for an HF model are:
1. Define the model `Model.register` annotation in a new `Model` subclass, example:
```python
@Model .register("MyModelForCausalLM")
class MyModel(Model):
model_arch = gguf.MODEL_ARCH.GROK
```
2024-07-15 13:46:39 +02:00
2. Define the layout of the GGUF tensors in [constants.py ](/gguf-py/gguf/constants.py )
2024-04-10 08:58:48 +02:00
Add an enum entry in `MODEL_ARCH` , the model human friendly name in `MODEL_ARCH_NAMES` and the GGUF tensor names in `MODEL_TENSORS` .
Example for `falcon` model:
```python
MODEL_ARCH.FALCON: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.OUTPUT_NORM,
MODEL_TENSOR.OUTPUT,
MODEL_TENSOR.ATTN_NORM,
MODEL_TENSOR.ATTN_NORM_2,
MODEL_TENSOR.ATTN_QKV,
MODEL_TENSOR.ATTN_OUT,
MODEL_TENSOR.FFN_DOWN,
MODEL_TENSOR.FFN_UP,
]
```
3. Map the original tensor names to the standardize equivalent in GGUF
As a general rule, before adding a new tensor name to GGUF, be sure the equivalent naming does not already exist.
2024-07-15 13:46:39 +02:00
Once you have found the GGUF tensor name equivalent, add it to the [tensor_mapping.py ](/gguf-py/gguf/tensor_mapping.py ) file.
2024-04-10 08:58:48 +02:00
If the tensor name is part of a repetitive layer/block, the key word `bid` substitutes it.
Example for the normalization tensor in attention layers:
```python
block_mappings_cfg: dict[MODEL_TENSOR, tuple[str, ...]] = {
# Attention norm
MODEL_TENSOR.ATTN_NORM: (
"gpt_neox.layers.{bid}.input_layernorm", # gptneox
"transformer.h.{bid}.ln_1", # gpt2 gpt-j refact qwen
"transformer.blocks.{bid}.norm_1", # mpt
...
)
}
```
`transformer.blocks.{bid}.norm_1` will be mapped to `blk.{bid}.attn_norm` in GGUF.
Depending on the model configuration, tokenizer, code and tensors layout, you will have to override:
- `Model#set_gguf_parameters`
- `Model#set_vocab`
- `Model#write_tensors`
NOTE: Tensor names must end with `.weight` suffix, that is the convention and several tools like `quantize` expect this to proceed the weights.
### 2. Define the model architecture in `llama.cpp`
The model params and tensors layout must be defined in `llama.cpp` :
1. Define a new `llm_arch`
2. Define the tensors layout in `LLM_TENSOR_NAMES`
3. Add any non standard metadata in `llm_load_hparams`
4. Create the tensors for inference in `llm_load_tensors`
5. If the model has a RoPE operation, add the rope type in `llama_rope_type`
NOTE: The dimensions in `ggml` are typically in the reverse order of the `pytorch` dimensions.
### 3. Build the GGML graph implementation
This is the funniest part, you have to provide the inference graph implementation of the new model architecture in `llama_build_graph` .
2024-05-07 17:20:33 +02:00
Have a look at existing implementation like `build_llama` , `build_dbrx` or `build_bert` .
2024-04-10 08:58:48 +02:00
2024-05-07 17:20:33 +02:00
When implementing a new graph, please note that the underlying `ggml` backends might not support them all, support for missing backend operations can be added in another PR.
2024-04-10 08:58:48 +02:00
2024-07-15 13:46:39 +02:00
Note: to debug the inference graph: you can use [llama-eval-callback ](/examples/eval-callback/ ).
2024-04-11 14:51:07 +02:00
2024-04-10 08:58:48 +02:00
## GGUF specification
https://github.com/ggerganov/ggml/blob/master/docs/gguf.md
## Resources
- YaRN RoPE scaling https://github.com/ggerganov/llama.cpp/pull/2268
- support Baichuan serial models https://github.com/ggerganov/llama.cpp/pull/3009
- support attention bias https://github.com/ggerganov/llama.cpp/pull/4283
- Mixtral support https://github.com/ggerganov/llama.cpp/pull/4406
- BERT embeddings https://github.com/ggerganov/llama.cpp/pull/5423
- Grok-1 support https://github.com/ggerganov/llama.cpp/pull/6204
- Command R Plus support https://github.com/ggerganov/llama.cpp/pull/6491
- support arch DBRX https://github.com/ggerganov/llama.cpp/pull/6515
- How to convert HuggingFace model to GGUF format https://github.com/ggerganov/llama.cpp/discussions/2948