HebArabNlpProject/Hebatron

Full fine-tune vs nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16. Fine-tune: 100% of the weights changed, across 52 of 52 layers.

Tensors changed: other 207/207, Embeddings 1/1, Norms 76/76, MLP 2691/2691, Attention 24/24, lm_head 1/1

2 derivatives checked

• mradermacher/Hebatron-i1-GGUF: Quantization
• mradermacher/Hebatron-GGUF: Quantization

1 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse