Modelpublic

hannah-khallaf/e2r-mistral-7b-qlora-merged7

sourceHugging Faceupdated 3mo agoView on Hugging Face

0likes3downloads

Model Card

e2r-mistral-7b-qlora-merged7

Portable PEFT release of the selected merged-seven QLoRA checkpoint.

—Selected seed: 42
—Base model: mistralai/Mistral-7B-Instruct-v0.3
—PEFT task type: CAUSAL_LM
—Number of strategy labels: 7

Loading

Load the base model with AutoModelForCausalLM, then attach the adapter with PeftModel.from_pretrained.

The adapter is not a SequenceClassification head. Predictions must use the same prompt format and label parsing used during evaluation. See resolved_config.yaml and inference_config.json.