boradorish/llama3-3B-sft

TEXT GENERATIONConcurrency Cost:1Model Size:3.2BQuant:BF16Ctx Length:32kTool Calling:SupportedPublished:May 25, 2026License:otherArchitecture:Transformer Cold

The boradorish/llama3-3B-sft model is a 3.2 billion parameter instruction-tuned causal language model, fine-tuned by boradorish from Meta's Llama-3.2-3B-Instruct. This model specializes in reasoning tasks, having been fine-tuned on the sunny_reasoning dataset. It offers a 32768-token context length, making it suitable for applications requiring detailed contextual understanding in reasoning-focused scenarios.

Loading preview...

Overview

This model, boradorish/llama3-3B-sft, is a fine-tuned variant of Meta's Llama-3.2-3B-Instruct, specifically optimized for reasoning tasks. With 3.2 billion parameters and a substantial 32768-token context length, it aims to provide enhanced performance in scenarios requiring logical deduction and problem-solving.

Key Characteristics

  • Base Model: Fine-tuned from meta-llama/Llama-3.2-3B-Instruct.
  • Parameter Count: 3.2 billion parameters.
  • Context Length: Supports a 32768-token context window.
  • Specialization: Fine-tuned on the sunny_reasoning dataset, indicating a focus on improving reasoning capabilities.

Training Details

The model was trained using specific hyperparameters to achieve its specialized performance:

  • Learning Rate: 4e-05
  • Batch Size: A total training batch size of 64 (4 per device with 8 gradient accumulation steps on 2 GPUs).
  • Optimizer: ADAMW_TORCH_FUSED with betas=(0.9, 0.999) and epsilon=1e-08.
  • Scheduler: Cosine learning rate scheduler with a 0.1 warmup ratio over 3 epochs.

Intended Use Cases

Given its fine-tuning on a reasoning dataset, this model is particularly well-suited for applications that demand strong logical inference and problem-solving abilities. Developers looking for a compact yet capable model for reasoning-intensive tasks should consider this offering.