CreitinGameplays/Llama-3.1-8B-R1-v0.1

Llama 3.1 8B R1 v0.1

Took 28 hours to finetune on 2x Nvidia RTX A6000 with the following settings:

• Batch size: 8
• Gradient accumulation steps: 1
• Epochs: 2
• Learning rate: 1e-4
• Warmup ratio: 0.1

Run the model:

import torch
from transformers import pipeline

model_id = "CreitinGameplays/Llama-3.1-8B-R1-v0.1"

pipe = pipeline(
    "text-generation",
    model=model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto"
)

messages = [
    {"role": "system", "content": "You are an AI assistant named Llama, made by Meta AI."},
    {"role": "user", "content": "How many r's are in strawberry?"}
]

outputs = pipe(
    messages,
    temperature=0.6,
    repetition_penalty=1.1,
    max_new_tokens=2048
)

print(outputs[0]["generated_text"][-1])

Current Limitations

The model may not output the final response after the reasoning step.