harshbhatt7585/tinygroot-hrm-loop-pdr12-sft-1000-plus500

tinyGroot checkpoint uploaded from the training pipeline.

Checkpoint

• Step: 500
• Tokenizer: nanochat
• Layers: 12
• Hidden size: 768
• Heads: 6
• MTP heads: 1
• Max sequence length: 2048
• Source run: hrm-loop-pdr12-sft-1000-plus500

Files

• model.pt: model state_dict (mmap+weights_only=True friendly).
• optimizer.pt: optimizer and grad-scaler state (only needed to resume training).
• meta.json: step, model config, tokenizer type, and original training args.
• tokenizer_hf/tokenizer.json: tokenizer used by this checkpoint.

Load this checkpoint with the tinyGroot codebase, not the Transformers AutoModel API.