Files
ModelHub XC 191f715433 初始化项目,由ModelHub XC社区提供模型
Model: starVLA/Qwen3-VL-4B-Instruct-Action
Source: Original Platform
2026-08-16 12:47:13 +08:00

9 lines
480 B
Markdown

The key difference is that Qwen3-VL-4B-Instruct-Action extends the original tokenizer with an additional 2048 action tokens. Otherwise, it is identical to Qwen3-VL-4B-Instruct—no additional fine-tuning was applied.
To add these tokens, we used the script:
starVLA/model/modules/vlm/tools/add_qwen_special_tokens
The model simply expands the token codebook to accommodate action token embeddings, but the underlying VLM weights remain unchanged from the base Instruct model.