minsu0567/IAD-X1-SFT-answer-last

VISIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:4.5BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Jun 27, 2026License:otherArchitecture:Transformer Featherless Exclusive Cold

minsu0567/IAD-X1-SFT-answer-last is a fine-tuned version of the Qwen3.5-4B causal language model, developed by Qwen. This model has been specifically fine-tuned on the PA_SFT_2_answer_last dataset, indicating an optimization for tasks requiring specific answer extraction or generation based on the 'answer_last' methodology. It is designed for applications where precise, context-aware responses are critical, leveraging the Qwen3.5 architecture for enhanced performance in targeted question-answering scenarios.

Loading preview...

Model Overview

This model, minsu0567/IAD-X1-SFT-answer-last, is a specialized fine-tuned variant of the Qwen3.5-4B base model, originally developed by Qwen. It has undergone supervised fine-tuning (SFT) using the PA_SFT_2_answer_last dataset, suggesting an optimization for tasks where the final answer is a critical component of the output.

Key Characteristics

  • Base Model: Qwen3.5-4B, a robust causal language model.
  • Fine-tuning Objective: Optimized for specific answer generation, likely in question-answering or response-focused tasks, as indicated by the answer_last dataset.
  • Training Details:
    • Learning Rate: 1e-05
    • Optimizer: ADAMW_BNB
    • Epochs: 1.0
    • Batch Size: 1 (train), 8 (eval) with 2 gradient accumulation steps.

Intended Use Cases

This model is particularly suited for applications requiring precise and contextually relevant answers, especially in scenarios where the 'answer_last' format or methodology is beneficial. Developers might consider this model for:

  • Targeted Question Answering: Generating concise answers from provided contexts.
  • Information Extraction: Extracting specific pieces of information that typically appear as the final answer in a sequence.
  • Automated Response Generation: Creating focused responses in conversational AI or support systems where a direct answer is expected.