tkwiecinski/amr-fma-Qwen3-8B-lora_sft-block_em_auto_good-e7_qwen3_blockem-s42

amr-fma training run.

• Method: lora_sft
• Base model: Qwen/Qwen3-8B
• Dataset: auto_correct (slug: block_em_auto_good)
• Seed: 42
• Git commit: 94a4b55454f8dbc4a1f8226c2f7035cb46a3938f
• Exp name: e7_qwen3_blockem
• WandB run: w9d4tkpm

Tags

• phase:P1
• domain:automotive

Checkpoints (branches)

• step 1 → revision step-00001
• step 2 → revision step-00002
• step 5 → revision step-00005
• step 8 → revision step-00008
• step 15 → revision step-00015
• step 26 → revision step-00026
• step 45 → revision step-00045

Pin a specific checkpoint with revision=... in AutoModelForCausalLM.from_pretrained / PeftModel.from_pretrained.

Hyperparameter sections

checkpointing, dataset, evaluation, final_adapter_path, lora, model, optimization, prompt_style, runtime, sdpo, sequence, total_steps