---
library_name: peft
base_model: meta-llama/Llama-3.1-8B-Instruct
tags:
- amr-fma
- lora_sft
- domain:code
- phase:P1
---

# tkwiecinski/amr-fma-Llama-3.1-8B-Instruct-lora_sft-em_secure-e2_em_ablation-s43

amr-fma training run.

- **Method**: `lora_sft`
- **Base model**: `meta-llama/Llama-3.1-8B-Instruct`
- **Dataset**: `secure` (slug: `em_secure`)
- **Seed**: `43`
- **Git commit**: `6b9a7d1eb9d81484964a71ad6829945e7680efad`
- **Exp name**: `e2_em_ablation`
- **WandB run**: `u1nhw7f4`

## Tags
- phase:P1
- domain:code

## Checkpoints (branches)
- step 1 → revision `step-00001`
- step 3 → revision `step-00003`
- step 5 → revision `step-00005`
- step 10 → revision `step-00010`
- step 17 → revision `step-00017`
- step 31 → revision `step-00031`
- step 57 → revision `step-00057`

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`