exp-0223-027-realobs-llmagent-qwen2.5-7b/README.md

base_model, datasets, language, license, library_name, pipeline_tag, tags

base_model	datasets	language	license	library_name	pipeline_tag	tags
Qwen/Qwen2.5-7B-Instruct	Yano/exp-0223-027-realobs-llmagent-alfworld-data	en	apache-2.0	transformers	text-generation	qlora lora merged alfworld agent realobs-llmagent

exp-0223-027-realobs-llmagent-qwen2.5-7b

Fine-tuned from Qwen/Qwen2.5-7B-Instruct using QLoRA (4-bit, Unsloth).

Purpose

ALFWorld SFT combining real environment observations from experiment 016 with LLM-renarrated agent responses (strategic THOUGHTs + ACTION-dominant format).

Key Design

• Environment responses (user): Real data from 016 (byte-exact preservation)
• Agent responses (assistant): LLM-renarrated (strategic THOUGHT, not template)
• Format: First turn THOUGHT+ACTION, subsequent ACTION-only, recovery THOUGHT after failures
• Failure patterns: Naturally inherited from 016's 72% failure-containing trajectories

Training Configuration

• Base model: Qwen/Qwen2.5-7B-Instruct
• Method: QLoRA (4-bit), merged to 16-bit
• Max sequence length: 2048
• Epochs: 3
• Learning rate: 2e-05
• LoRA: r=64, alpha=128
• Collator: AllAssistantTurnsCollator (all turns supervised)