Model Overview

This model, exp-uns-r2egym-33_6x_glm_4_7_traces_jupiter_cleaned, is an 8 billion parameter language model developed by laion. It is a fine-tuned version of the Qwen/Qwen3-8B base model, adapted for specific applications through further training.

Key Characteristics

• Base Model: Qwen/Qwen3-8B
• Parameter Count: 8 billion parameters
• Context Length: 32768 tokens
• Fine-tuning Dataset: The model was fine-tuned on the /data/cat/ws/befe330h-befe330h-otagent/huggingface/hub/datasets--DCAgent--exp-uns-r2egym-33_6x_glm_4.7_traces_jupiter_cleaned/snapshots/979a62e8e35d7341b236ceb175f7ddcff8c72d01_thinking_preprocessed dataset.

Training Details

The fine-tuning process utilized the following hyperparameters:

• Learning Rate: 4e-05
• Optimizer: ADAMW_TORCH_FUSED with betas=(0.9, 0.98) and epsilon=1e-08
• Batch Size: A total training batch size of 16 (1 per device, 8 devices, 2 gradient accumulation steps)
• Epochs: 7.0
• Scheduler: Cosine learning rate scheduler with 0.1 warmup ratio

Intended Use

Given its fine-tuning on a specific dataset related to 'traces' and 'jupiter_cleaned', this model is likely intended for tasks that align with the characteristics and domain of its training data. Users should evaluate its performance on their specific use cases, particularly those involving trace analysis or similar specialized reasoning tasks.