vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

llama : add Mixtral support (#4406)

Browse source 
* convert : support Mixtral as LLAMA arch

* convert : fix n_ff typo

* llama : model loading

* ggml : sync latest ggml_mul_mat_id

* llama : update graph to support MoE

* llama : fix cur -> cur_expert

* llama : first working version

* llama : fix expert weighting in the FFN

* ggml : ggml_get_rows support 2D indexing [n_tokens, n_experts] (cpu only)

* ggml : add n_as argument to ggml_mul_mat_id

* ggml : fix ggml_get_rows to take into account ne02 / ne11

* metal : add more general support for ggml_get_rows + tests

* llama : add basic support for offloading moe with CUDA

* metal : add/mul/div use general kernel when src1 not cont

* metal : reduce the kernel launches for ggml_mul_mat_id

* ggml : get_rows : support non-contiguos tensors with gaps, generalize up to 3D

* ggml : update get_rows f16 and q

* cuda : support non-contiguous src1 in get_rows

* llama : offload missing ffn_moe_silu

* metal : fix ggml_get_rows to work with non-cont src1

* metal : add indirect mat-vec kernels for all quantization types

* llama : do not quantize expert gating tensors

* llama : add n_expert and n_expert_used to hparams + change quants

* test-backend-ops : add moe test

* cuda : fix get_rows when ncols is odd

* convert : determine n_ctx correctly

* metal : fix ggml_mul_mat_id for F32

* test-backend-ops : make experts more evenly probable (test_moe)

* test-backend-ops : cleanup, add moe test for batches

* test-backend-ops : add cpy from f32 -> all types test

* test-backend-ops : fix dequantize block offset

* llama : fix hard-coded number of experts

* test-backend-ops : simplify and disable slow tests to avoid CI timeout

* test-backend-ops : disable MOE test with thread sanitizer

* cuda : fix mul_mat_id with multi gpu

* convert : use 1e6 rope_freq_base for mixtral

* convert : fix style

* convert : support safetensors format

* gguf-py : bump version

* metal : add cpy f16 -> f32 kernel

* metal : fix binary ops for ne10 % 4 != 0

* test-backend-ops : add one more sum_rows test

* ggml : do not use BLAS with ggml_mul_mat_id

* convert-hf : support for mixtral-instruct (#4428)

* convert : typo fix, add additional hyperparameters, use LLaMA arch for Mixtral-instruct

* convert : use sentencepiece tokenizer for Mixtral-instruct

* convert : make flake8 happy

* metal : fix soft_max kernels

ref: 1914017863

* metal : limit kernels to not use more than the allowed threads

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
Co-authored-by: Radek Pilar <github@mrkva.eu>
This commit is contained in:
slaren 2023-12-13 13:04:25 +01:00 • committed by GitHub
parent fecac45658
commit 799a1cb13b
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
14 changed files with 2370 additions and 395 deletions
Download patch file Download diff file Expand all files Collapse all files

21
convert-hf-to-gguf.py
View file

@ -77,8 +77,18 @@ class Model:
self.gguf_writer.add_embedding_length(n_embd)
if (n_ff := self.hparams.get("intermediate_size")) is not None:
self.gguf_writer.add_feed_forward_length(n_ff)
if (n_head := self.hparams.get("num_attention_head")) is not None:
if (n_head := self.hparams.get("num_attention_heads")) is not None:
self.gguf_writer.add_head_count(n_head)
if (n_head_kv := self.hparams.get("num_key_value_heads")) is not None:
self.gguf_writer.add_head_count_kv(n_head_kv)
if (n_rms_eps := self.hparams.get("rms_norm_eps")) is not None:
self.gguf_writer.add_layer_norm_rms_eps(n_rms_eps)
if (n_experts := self.hparams.get("num_local_experts")) is not None:
self.gguf_writer.add_expert_count(n_experts)
if (n_experts_used := self.hparams.get("num_experts_per_tok")) is not None:
self.gguf_writer.add_expert_used_count(n_experts_used)
self.gguf_writer.add_parallel_residual(self.hparams.get("use_parallel_residual", True))
def write_tensors(self):
@ -170,6 +180,8 @@ class Model:
return StableLMModel
if model_architecture == "QWenLMHeadModel":
return QwenModel
if model_architecture == "MixtralForCausalLM":
return MixtralModel
return Model
def _is_model_safetensors(self) -> bool:
@ -207,6 +219,8 @@ class Model:
return gguf.MODEL_ARCH.STABLELM
if arch == "QWenLMHeadModel":
return gguf.MODEL_ARCH.QWEN
if arch == "MixtralForCausalLM":
return gguf.MODEL_ARCH.LLAMA
raise NotImplementedError(f'Architecture "{arch}" not supported!')
@ -837,6 +851,11 @@ class StableLMModel(Model):
self.gguf_writer.add_layer_norm_eps(1e-5)
class MixtralModel(Model):
def set_vocab(self):
self._set_vocab_sentencepiece()
class QwenModel(Model):
@staticmethod
def token_bytes_to_string(b):