Skip to content

fix: fix the bug of loading Internvl3 #8067

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
merged 6 commits into from
Jul 20, 2025
Merged

fix: fix the bug of loading Internvl3 #8067

Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
9b51175
bug fix of internvl loading
coco-alen Jul 15, 2025
b498e32
Merge branch 'main' into internvl-load-bugfix
jinleic Jul 16, 2025
ae27874
Merge branch 'main' into internvl-load-bugfix
ispobock Jul 17, 2025
c1c13de
Merge branch 'main' into internvl-load-bugfix
jinleic Jul 17, 2025
e7a9b43
fix lint
JustinTong0323 Jul 20, 2025
402cf8e
Merge branch 'main' into internvl-load-bugfix
JustinTong0323 Jul 20, 2025
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
3 changes: 3 additions & 0 deletions python/sglang/srt/configs/internvl.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,6 +9,7 @@
LlamaConfig,
PretrainedConfig,
PreTrainedTokenizer,
Qwen2Config,
)

from sglang.utils import logger
Expand Down Expand Up @@ -311,6 +312,8 @@ def __init__(
self.llm_config = LlamaConfig(**llm_config)
elif llm_config.get("architectures")[0] == "InternLM2ForCausalLM":
self.llm_config = InternLM2Config(**llm_config)
elif llm_config.get("architectures")[0] == "Qwen2ForCausalLM":
self.llm_config = Qwen2Config(**llm_config)
Comment on lines +315 to +316
Copy link
Contributor

@gemini-code-assist gemini-code-assist bot Jul 15, 2025

Choose a reason for hiding this comment

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

medium

While this change is correct, the overall conditional block could be improved for robustness and maintainability.

  • Robustness: The repeated use of llm_config.get("architectures")[0] is unsafe. It can raise a TypeError or IndexError if architectures is missing or empty. It's better to extract the architecture name into a variable with proper checks.
  • Maintainability: An if/elif chain for architectures is not easily scalable. A dictionary mapping architecture names to config classes would be a more maintainable pattern.

Here's a suggested refactoring for the whole block for future consideration:

LLM_CONFIG_MAPPING = {
    "LlamaForCausalLM": LlamaConfig,
    "InternLM2ForCausalLM": InternLM2Config,
    "Qwen2ForCausalLM": Qwen2Config,
}

architectures = llm_config.get("architectures")
if not architectures:
    raise ValueError("'architectures' key is missing or empty in llm_config.")

arch = architectures[0]
config_class = LLM_CONFIG_MAPPING.get(arch)

if config_class:
    self.llm_config = config_class(**llm_config)
else:
    raise ValueError(f"Unsupported architecture: {arch}")

else:
raise ValueError(
"Unsupported architecture: {}".format(
Expand Down
Loading