zhangyiss/mlx-examples
1
0
Fork 0
mirror of https://github.com/ml-explore/mlx-examples.git synced 2025-09-01 04:14:38 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

Add support for qwen2moe (#640)

Browse Source 
* add sparsemoe block and update decoder logic

* update file name to match HF

* update name

* Code formatting

* update gates calculation

* add support for Qwen2MoE.

* fix pytest

* code formatting and fix missing comma in utils

* Remove decoder sparse step.

Co-authored-by: bozheng-hit <dsoul0621@gmail.com>

* remove gate layer anti-quantisation

* remove unused argument

---------

Co-authored-by: bozheng-hit <dsoul0621@gmail.com>
This commit is contained in:
Prince Canuma
2024-04-02 20:33:29 +02:00
committed by GitHub
parent 78c431dc25
commit d661440dbb
3 changed files with 282 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
llms/mlx_lm/tuner/utils.py
View File

@@ -70,6 +70,7 @@ def linear_to_lora_layers(
"mixtral",
"stablelm",
"qwen2",
"qwen2_moe",
"gemma",
"starcoder2",
"cohere",
@@ -77,6 +78,9 @@ def linear_to_lora_layers(
keys = set(["self_attn.q_proj", "self_attn.v_proj"])
if model.model_type == "mixtral":
keys.add("block_sparse_moe.gate")
if model.model_type == "qwen2_moe":
keys.add("mlp.gate")
keys.add("mlp.shared_expert_gate")
elif model.model_type == "olmo":
keys = set(["att_proj"])
elif model.model_type == "phi-msft":