project-nils — project-nils-run-a

Training checkpoints for project-nils.

Base architecture: Qwen3.5-4B
RL stage, variant A. Contents: checkpoint-2900, checkpoint-2950, checkpoint-3000, final/. Each contains model.safetensors + DeepSpeed ZeRO shards.
Weights are model.safetensors per checkpoint; global_step*/ folders hold DeepSpeed ZeRO optimizer shards (only needed to resume training). Use zero_to_fp32.py to reconstruct fp32 weights from shards if needed.

No further details are provided.

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support