WithinUsAI/IBM-Grok4-Ultra.Fast.Coder-1B

Full fine-tune vs ibm-granite/granite-4.0-1b. Fine-tune: 100% of the weights changed, across 40 of 40 layers.

Tensors changed: Embeddings 1/1, Norms 33/81, Attention 160/160, MLP 80/80

2 derivatives checked

• mradermacher/IBM-Grok4-UltraFast-Coder-1B-GGUF: Same weights, converted
• WithinUsAI/IBM-Grok4-Ultra.Fast.Coder-1B-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse