423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2026-09-14 19:25:43 +08:00
Code Issues Packages Projects Releases Wiki Activity

[model] add Qwen3.8 model support (#10749)

Browse Source 
Co-authored-by: hiyouga <hiyouga@buaa.edu.cn>
This commit is contained in:
Hertz
2026-09-02 15:52:22 +08:00
committed by GitHub
parent d6bb97ddff
commit 4451765a6b
13 changed files with 272 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

33
src/llamafactory/extras/constants.py
View File

@@ -71,6 +71,8 @@ MCA_SUPPORTED_MODELS = {
"qwen3_next",
"qwen3_5",
"qwen3_5_moe",
"qwen3_5_moe_text",
"qwen3_5_text",
}
# Text LLM model_types supported by the Megatron Bridge PT/SFT path (gpt_step).
@@ -3000,6 +3002,37 @@ register_model_group(
)
register_model_group(
models={
"Qwen3.8-27B": {
DownloadSource.DEFAULT: "Qwen/Qwen3.8-27B",
DownloadSource.MODELSCOPE: "Qwen/Qwen3.8-27B",
},
"Qwen3.8-27B-FP8": {
DownloadSource.DEFAULT: "Qwen/Qwen3.8-27B-FP8",
DownloadSource.MODELSCOPE: "Qwen/Qwen3.8-27B-FP8",
},
},
template="qwen3_8",
multimodal=True,
)
register_model_group(
models={
"Qwen3.8-2.4T-A95B-Thinking": {
DownloadSource.DEFAULT: "Qwen/Qwen3.8-2.4T-A95B",
DownloadSource.MODELSCOPE: "Qwen/Qwen3.8-2.4T-A95B",
},
"Qwen3.8-2.4T-A95B-Thinking-FP8": {
DownloadSource.DEFAULT: "Qwen/Qwen3.8-2.4T-A95B-FP8",
DownloadSource.MODELSCOPE: "Qwen/Qwen3.8-2.4T-A95B-FP8",
},
},
template="qwen3_8",
)
register_model_group(
models={
"Qwen2-Audio-7B": {