trollek/ThoughtStream-4B-v0.2

Full fine-tune vs trollek/LittleInstructionMaker-4B-v0.2. Fine-tune: 100% of the weights changed, across 24 of 24 layers.

Tensors changed: Norms 31/49, MLP 72/72, Attention 96/96

2 derivatives checked

• Goekdeniz-Guelmez/ThoughtStream-4B-v0.2-gguf: Quantization
• mradermacher/ThoughtStream-4B-v0.2-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse