This repository contains a full merged Transformers checkpoint produced by applying the LoRA adapter from train_2026-05-22-19-31-02 onto the base model kuotient/Meta-Llama-3-8B-Instruct.

Base model

Base: kuotient/Meta-Llama-3-8B-Instruct
Fine-tuning framework: LLaMA-Factory
Template: llama3
Fine-tuning type: LoRA

Training summary

Dataset: simple_train
Epochs: 3.0
Learning rate: 5e-5
LoRA rank: 8
LoRA alpha: 16
Batch size per device: 2
Gradient accumulation steps: 8

Files

This repo is stored in standard Hugging Face Transformers format and includes:

config.json
generation_config.json
model-*.safetensors
model.safetensors.index.json
tokenizer files

Example

python

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

model_id = "cat-claws/Llama-3.1-8B-Instruct-lora-train_2026-05-22-19-31-02"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto",
)

messages = [
    {"role": "user", "content": "Who are you?"},
]
input_ids = tokenizer.apply_chat_template(
    messages,
    add_generation_prompt=True,
    return_tensors="pt",
).to(model.device)

outputs = model.generate(input_ids, max_new_tokens=256)
print(tokenizer.decode(outputs[0][input_ids.shape[-1]:], skip_special_tokens=True))