📚 The doc issue
Hello,
My collaborators and I are trying to compare minimal pair VLM LM models. But it is unclear what base model the InternVL3.5 model is using. The docs/paper says its "Qwen3-8B" but qwen changed the way they title models -- in the past QwenX-8B would mean it's the base model, with the instruct model being titled "QwenX-8B-Instruct". For Qwen3-8B they have "Qwen3-8B", which actually now is a instruct-tuned/thinking model and "Qwen3-8B-Base" is the non-thinking model. Please let us know which one is right (and if possible also clarify in the docs).
Thanks,
Kanishka
Suggest a potential alternative/fix
No response
📚 The doc issue
Hello,
My collaborators and I are trying to compare minimal pair VLM LM models. But it is unclear what base model the InternVL3.5 model is using. The docs/paper says its "Qwen3-8B" but qwen changed the way they title models -- in the past QwenX-8B would mean it's the base model, with the instruct model being titled "QwenX-8B-Instruct". For Qwen3-8B they have "Qwen3-8B", which actually now is a instruct-tuned/thinking model and "Qwen3-8B-Base" is the non-thinking model. Please let us know which one is right (and if possible also clarify in the docs).
Thanks,
Kanishka
Suggest a potential alternative/fix
No response