adamo1139's picture
Update README.md
924cf1e verified
metadata
license: apache-2.0
base_model:
  - Qwen/Qwen2-VL-7B-Instruct

Qwen2-VL-7B-Instruct finetuned on 60000 first samples from Llava-Instruct-150k.json dataset via LoRA.

Prompt format is the same. It's responses are somewhat shorter while still accurate, it has less soap-ey vibe to it then Qwen2-7B-Instruct has but difference is minor.