Skip to content

feat: internvl llm support qwen3moe #947

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
hiworldwzj merged 1 commit into from
Jul 1, 2025
Merged

feat: internvl llm support qwen3moe #947

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
25 changes: 25 additions & 0 deletions lightllm/models/internvl/model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,6 +10,7 @@
from lightllm.models.phi3.model import Phi3TpPartModel
from lightllm.models.qwen2.model import Qwen2TpPartModel
from lightllm.models.qwen3.model import Qwen3TpPartModel
from lightllm.models.qwen3_moe.model import Qwen3MOEModel
from lightllm.models.deepseek2.model import Deepseek2TpPartModel
from lightllm.models.qwen_vl.layer_infer.pre_layer_infer import LlamaMultimodalPreLayerInfer
from lightllm.models.internvl.layer_weights.pre_and_post_layer_weight import (
Expand Down Expand Up @@ -297,3 +298,27 @@ def _init_config(self):
if self.finetune_config:
self.config["vocab_size"] = self.finetune_config.vocab_size
return


@ModelRegistry(["internvl_chat"], is_multimodal=True, condition=llm_model_type_is("qwen3_moe"))
class InternVLQwen3MOETpPartModel(Qwen3MOEModel):
# weight class
pre_and_post_weight_class = InternVLLlamaPreAndPostLayerWeight

# infer class
pre_layer_infer_class = LlamaMultimodalPreLayerInfer

def __init__(self, kvargs):
super().__init__(kvargs)
return

def _init_config(self):
with open(os.path.join(self.weight_dir_, "config.json"), "r") as json_file:
self.config = json.load(json_file)["llm_config"]
Comment on lines +316 to +317
Copy link
Contributor

@gemini-code-assist gemini-code-assist bot Jun 30, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

It's a good practice to explicitly specify the file encoding when opening text files to avoid potential UnicodeDecodeError on some platforms. Please add encoding="utf-8" to the open() call to ensure consistent behavior across different environments.

with open(os.path.join(self.weight_dir_, "config.json"), "r", encoding="utf-8") as json_file:
    self.config = json.load(json_file)["llm_config"]

# rename keys
repair_config(self.config, same_names=["num_attention_heads", "n_head"])
repair_config(self.config, same_names=["hidden_size", "n_embd", "n_embed"])
repair_config(self.config, same_names=["num_hidden_layers", "n_layer"])
if self.finetune_config:
self.config["vocab_size"] = self.finetune_config.vocab_size
return