423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-11-04 18:02:19 +08:00
Code Issues Packages Projects Releases Wiki Activity

[model] add models (#7054)

Browse Source 
* add qwen25vl awq models

* add moonlight

Former-commit-id: ae3be2970fea8a35907202a313ab767381c44916
This commit is contained in:
hoshi-hiyouga 2025-02-24 22:05:13 +08:00 committed by GitHub
parent ee46011b34
commit c1d5073bd3
7 changed files with 44 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
examples/train_full/llama3_full_sft.yaml
View File

@ -34,6 +34,7 @@ bf16: true
ddp_timeout: 180000000
### eval
# eval_dataset: alpaca_en_demo
# val_size: 0.1
# per_device_eval_batch_size: 1
# eval_strategy: steps

1
examples/train_lora/llama3_lora_dpo.yaml
View File

@ -37,6 +37,7 @@ bf16: true
ddp_timeout: 180000000
### eval
# eval_dataset: dpo_en_demo
# val_size: 0.1
# per_device_eval_batch_size: 1
# eval_strategy: steps

1
examples/train_lora/llama3_lora_pretrain.yaml
View File

@ -34,6 +34,7 @@ bf16: true
ddp_timeout: 180000000
### eval
# eval_dataset: c4_demo
# val_size: 0.1
# per_device_eval_batch_size: 1
# eval_strategy: steps

1
examples/train_lora/llama3_lora_reward.yaml
View File

@ -35,6 +35,7 @@ bf16: true
ddp_timeout: 180000000
### eval
# eval_dataset: dpo_en_demo
# val_size: 0.1
# per_device_eval_batch_size: 1
# eval_strategy: steps

1
examples/train_lora/llama3_lora_sft.yaml
View File

@ -35,6 +35,7 @@ bf16: true
ddp_timeout: 180000000
### eval
# eval_dataset: alpaca_en_demo
# val_size: 0.1
# per_device_eval_batch_size: 1
# eval_strategy: steps

12
src/llamafactory/data/template.py
View File

@ -985,6 +985,18 @@ register_template(
)
register_template(
name="moonlight",
format_user=StringFormatter(
slots=["<|im_user|>user<|im_middle|>{{content}}<|im_end|><|im_assistant|>assistant<|im_middle|>"]
),
format_assistant=StringFormatter(slots=["{{content}}<|im_end|>"]),
format_system=StringFormatter(slots=["<|im_system|>system<|im_middle|>{{content}}<|im_end|>"]),
default_system="You are a helpful assistant provided by Moonshot-AI.",
stop_words=["<|im_end|>"],
)
# copied from vicuna template
register_template(
name="llava",

27
src/llamafactory/extras/constants.py
View File

@ -1315,6 +1315,21 @@ register_model_group(
)
register_model_group(
models={
"Moonlight-16B-A3B": {
DownloadSource.DEFAULT: "moonshotai/Moonlight-16B-A3B",
DownloadSource.MODELSCOPE: "moonshotai/Moonlight-16B-A3B",
},
"Moonlight-16B-A3B-Instruct": {
DownloadSource.DEFAULT: "moonshotai/Moonlight-16B-A3B-Instruct",
DownloadSource.MODELSCOPE: "moonshotai/Moonlight-16B-A3B-Instruct",
},
},
template="moonlight",
)
register_model_group(
models={
"OLMo-1B": {
@ -2234,6 +2249,18 @@ register_model_group(
DownloadSource.DEFAULT: "Qwen/Qwen2.5-VL-72B-Instruct",
DownloadSource.MODELSCOPE: "Qwen/Qwen2.5-VL-72B-Instruct",
},
"Qwen2.5-VL-3B-Instruct-AWQ": {
DownloadSource.DEFAULT: "Qwen/Qwen2.5-VL-3B-Instruct-AWQ",
DownloadSource.MODELSCOPE: "Qwen/Qwen2.5-VL-3B-Instruct-AWQ",
},
"Qwen2.5-VL-7B-Instruct-AWQ": {
DownloadSource.DEFAULT: "Qwen/Qwen2.5-VL-7B-Instruct-AWQ",
DownloadSource.MODELSCOPE: "Qwen/Qwen2.5-VL-7B-Instruct-AWQ",
},
"Qwen2.5-VL-72B-Instruct-AWQ": {
DownloadSource.DEFAULT: "Qwen/Qwen2.5-VL-72B-Instruct-AWQ",
DownloadSource.MODELSCOPE: "Qwen/Qwen2.5-VL-72B-Instruct-AWQ",
},
},
template="qwen2_vl",
multimodal=True,