le-harnais-ft-agentworld-1b/README.md

license, base_model, library_name, tags

license	base_model	library_name	tags
llama3.2	meta-llama/Llama-3.2-1B-Instruct	transformers	le-harnais hero

le-harnais / ft-agentworld-1b

World-model student distilled from Qwen-AgentWorld-35B; Mac-runnable tool-outcome predictor.

• Base model: meta-llama/Llama-3.2-1B-Instruct — Built with Llama; Llama Community License applies.
• Class: hero
• Training data: datasets/agentworld_distill_train.jsonl (160 ex; teacher=Qwen-AgentWorld-35B-A3B)
• Headline: token-F1 0.826 / OBSERVATION hit-rate 60% vs teacher (n=40)

Reproduce

BASE=meta-llama/Llama-3.2-1B-Instruct OUT=refs/llm-jepa/ft-agentworld-1b tools/distill_agentworld.sh

Full recipe, datasets, and eval commands: see docs/REPRODUCE.md in the [le-harnais distribution]. Provenance & license: docs/PROVENANCE.md.

Formats in this repo

• *.safetensors — bf16 inference weights (serve with transformers or le-harnais lh-serve/candle).
• *.Q4_K_M.gguf — portable 4-bit quant (run via ollama / llama.cpp; Mac-friendly).
• *.Q8_0.gguf — higher-fidelity 8-bit quant (hero models).

Orchestration amplifies a capable generator; it does not create competence.