DCAgent/d1_mixed_original_swe_hardened_tb2_glm47

Model Overview

DCAgent/d1_mixed_original_swe_hardened_tb2_glm47 is an 8 billion parameter language model, fine-tuned from the robust Qwen/Qwen3-8B architecture. This model has been specifically trained on the /e/scratch/jureap59/raoof1/sft_data/hf_hub/datasets--DCAgent--d1_mixed_original_swe_hardened_tb2_glm47_traces dataset, which consists of "thinking preprocessed" traces.

Key Characteristics

• Base Model: Qwen/Qwen3-8B, known for its strong general language understanding.
• Specialized Fine-tuning: The training on "thinking preprocessed" traces indicates a focus on enhancing the model's ability to process and generate structured thought sequences, potentially improving its reasoning and problem-solving capabilities.
• Parameter Count: 8 billion parameters, offering a balance between performance and computational efficiency.
• Context Length: Supports a context length of 32768 tokens, allowing for processing extensive inputs and maintaining coherence over long interactions.

Training Details

The model was trained with a learning rate of 4e-05 over 7 epochs, utilizing a distributed setup across 16 devices. An AdamW optimizer with specific beta and epsilon values was employed, alongside a cosine learning rate scheduler with a 0.1 warmup ratio.

Potential Use Cases

Given its specialized training, this model is likely well-suited for applications requiring:

• Complex Reasoning: Tasks that benefit from a model's ability to follow and generate logical steps.
• Problem Solving: Scenarios where structured thinking and trace analysis are beneficial.
• Agentic Workflows: Potentially useful in environments where an AI agent needs to articulate its thought process or plan actions.