423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-08-23 06:12:50 +08:00
Code Issues Packages Projects Releases Wiki Activity

Tiny fix

Browse Source 
Former-commit-id: 8f13a3627d06a6f0a9b4e35443a415958d9ad1c9
This commit is contained in:
Kingsley 2024-09-29 00:00:23 +08:00
parent fd79cf8551
commit a2452d0b1c
5 changed files with 6 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
README.md
View File

@ -183,7 +183,7 @@ Compared to ChatGLM's [P-Tuning](https://github.com/THUDM/ChatGLM2-6B/tree/main/
| [PaliGemma](https://huggingface.co/google) | 3B | paligemma |
| [Phi-1.5/Phi-2](https://huggingface.co/microsoft) | 1.3B/2.7B | - |
| [Phi-3](https://huggingface.co/microsoft) | 4B/7B/14B | phi |
| [Pixtral](https://huggingface.co/mistralai/Pixtral-12B-2409) | 12B | pixtral |
| [Pixtral](https://huggingface.co/mistral-community/pixtral-12b) | 12B | pixtral |
| [Qwen (1-2.5) (Code/Math/MoE)](https://huggingface.co/Qwen) | 0.5B/1.5B/3B/7B/14B/32B/72B/110B | qwen |
| [Qwen2-VL](https://huggingface.co/Qwen) | 2B/7B/72B | qwen2_vl |
| [StarCoder 2](https://huggingface.co/bigcode) | 3B/7B/15B | - |

2
README_zh.md
View File

@ -184,7 +184,7 @@ https://github.com/user-attachments/assets/e6ce34b0-52d5-4f3e-a830-592106c4c272
| [PaliGemma](https://huggingface.co/google) | 3B | paligemma |
| [Phi-1.5/Phi-2](https://huggingface.co/microsoft) | 1.3B/2.7B | - |
| [Phi-3](https://huggingface.co/microsoft) | 4B/7B/14B | phi |
| [Pixtral](https://huggingface.co/mistralai/Pixtral-12B-2409) | 12B | pixtral |
| [Pixtral](https://huggingface.co/mistral-community/pixtral-12b) | 12B | pixtral |
| [Qwen (1-2.5) (Code/Math/MoE)](https://huggingface.co/Qwen) | 0.5B/1.5B/3B/7B/14B/32B/72B/110B | qwen |
| [Qwen2-VL](https://huggingface.co/Qwen) | 2B/7B/72B | qwen2_vl |
| [StarCoder 2](https://huggingface.co/bigcode) | 3B/7B/15B | - |

3
src/llamafactory/data/mm_plugin.py
View File

@ -24,7 +24,6 @@ if TYPE_CHECKING:
from av.stream import Stream
from transformers import PreTrainedTokenizer, ProcessorMixin
from transformers.image_processing_utils import BaseImageProcessor
from transformers.processing_utils import _validate_images_text_input_order, ProcessingKwargs
class EncodedImage(TypedDict):
path: Optional[str]
@ -392,7 +391,7 @@ class PixtralPlugin(BasePlugin):
mm_inputs = self._get_mm_inputs(images, videos, processor)
if mm_inputs.get("image_sizes"):
mm_inputs.pop("image_sizes")
if isinstance(mm_inputs.get("pixel_values"), list) and len(mm_inputs.get("pixel_values")[0]) >= 2:
raise ValueError("Now it only supports batchsize=1 on per gpu due to `List[tensor]` can not pack into BachEncoding")

6
src/llamafactory/extras/constants.py
View File

@ -1059,9 +1059,9 @@ register_model_group(
register_model_group(
models={
"Pixtral-12B-2409": {
DownloadSource.DEFAULT: "mistralai/Pixtral-12B-2409",
DownloadSource.MODELSCOPE: "LLM-Research/Pixtral-12B-2409",
"Pixtral-12B": {
DownloadSource.DEFAULT: "mistral-community/pixtral-12b",
DownloadSource.MODELSCOPE: "AI-ModelScope/pixtral-12b",
}
},
template="mistral",

2
src/llamafactory/model/model_utils/visual.py
View File

@ -96,7 +96,6 @@ def autocast_projector_dtype(model: "PreTrainedModel", model_args: "ModelArgumen
mm_projector: "torch.nn.Module" = getattr(model, "multi_modal_projector")
elif model_type == "qwen2_vl":
mm_projector: "torch.nn.Module" = getattr(getattr(model, "visual"), "merger")
# TODO check it
elif model_type == "pixtral":
mm_projector: "torch.nn.Module" = getattr(model, "vision_language_adapte")
else:
@ -128,7 +127,6 @@ def get_forbidden_modules(config: "PretrainedConfig", finetuning_args: "Finetuni
if model_type in ["llava", "paligemma", "pixtral"]:
if finetuning_args.freeze_vision_tower:
forbidden_modules.add("vision_tower")
#TODO check it
forbidden_modules.add("vision_encoder")
if finetuning_args.train_mm_proj_only: