SciWorld SFT skill modules (Qwen3-8B)

Nine task-family LoRA adapters for ScienceWorld from From History to State: Constant-Context Skill Learning for LLM Agents (NeurIPS 2026).

Each sub-folder is one adapter for Qwen/Qwen3-8B: find_X, lifespan, conductivity, measurement, growth, state_change, power, chemistry, life_stages.

Code and evaluation instructions: https://github.com/haoyangxie/constant-context-skill-learning

Downloads last month
-
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for ricemonster/sciworld_sft_all

Finetuned
Qwen/Qwen3-8B
Adapter
(2415)
this model