CompassioninMachineLearning/Qwen3-8b-urban-seed2-20261012-CPT-best-val-step-750
TEXT GENERATIONPricing:Input $0.468 / Output $1.82Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Oct 13, 2026Architecture:Transformer Featherless Exclusive Cold
The CompassioninMachineLearning/Qwen3-8b-urban-seed2-20261012-CPT-best-val-step-750 is an 8 billion parameter Qwen3-based language model developed by CompassioninMachineLearning. This model is a standalone merged BF16 model, trained on the `urban_12738_cleaned` dataset with 10,072 distinct documents and 2,000 repeat exposures per epoch. It is designed for tasks related to urban data, with weights packaged losslessly into eight safetensors shards, requiring no adapter for loading.
Loading preview...
Overview
This model, CompassioninMachineLearning/Qwen3-8b-urban-seed2-20261012-CPT-best-val-step-750, is an 8 billion parameter Qwen3-based language model. It is a standalone merged BF16 model, derived from epoch 1.9860834990059644 at step 750.
Key Capabilities
- Urban Data Specialization: Trained on the
CompassioninMachineLearning/urban_12738_cleaneddataset, indicating a focus on processing and understanding urban-related information. - Efficient Loading: Merged using Unsloth's native
save_pretrained_merged(save_method="merged_16bit"), allowing it to be loaded without an adapter. - Robust Weights: The weights are validated BF16 and packaged losslessly into eight safetensors shards, ensuring data integrity.
Good for
- Urban-centric NLP tasks: Ideal for applications requiring analysis or generation of text related to urban environments, planning, or social dynamics.
- Research and Development: Suitable for researchers exploring fine-tuning or domain adaptation on specific datasets, particularly those with urban contexts.
- Deployment without Adapters: Simplifies deployment workflows as it does not require an adapter to load, making it easier to integrate into existing systems.