SlowGuess/ABForge-Qwen3-8B-Task1-SFT

TEXT GENERATIONPricing:Input $0.468 / Output $1.82Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jun 11, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The SlowGuess/ABForge-Qwen3-8B-Task1-SFT model is an 8 billion parameter language model, supervised fine-tuned from Qwen/Qwen3-8B. Developed by SlowGuess, it is specifically designed for "Ablation Objective Identification" within the ABForge framework. This model excels at proposing candidate ablation objectives, identifying a target module and a research question, based on the ablation-free context of research papers. Its primary strength lies in automating the design of scientific ablations for research analysis.

Loading preview...

ABForge-Qwen3-8B-Task1-SFT: Ablation Objective Identification

This model, developed by SlowGuess, is an 8 billion parameter language model (Qwen3-8B) that has undergone supervised fine-tuning (SFT) as part of the ABForge post-training pipeline. Its core function is to assist in Task 1: Ablation Objective Identification.

Key Capabilities

  • Proposes Ablation Objectives: Given the ablation-free context of a research paper, the model generates potential ablation objectives.
  • Identifies Target Modules: For each proposed objective, it specifies the component or module intended for ablation.
  • Formulates Research Questions: It pairs each target module with a research question that the ablation is designed to answer.
  • Paper-Grounded Design: The model's outputs are grounded in the provided research paper context, facilitating systematic ablation design.

Training and Evaluation

The model was SFT on sft_task1_45961.jsonl from the SlowGuess/abforge-data dataset, which is derived from CC-licensed research papers. Evaluation is performed using the held-out AblationBench split of the same dataset. Users can reproduce evaluation using the SlowGuess/Abforge_1 code.

Good For

  • Researchers and scientists looking to automate or streamline the design of ablation studies.
  • Generating structured ablation objectives (Target Module + Research Question) from research paper text.
  • Understanding the ABForge framework for paper-grounded ablation design.