POLARIS-Project/Polaris-1.7B-Preview
Full fine-tune vs Qwen/Qwen3-1.7B. Fine-tune: 100% of the weights changed, across 28 of 28 layers.
Tensors changed: Embeddings 1/1, Norms 89/113, MLP 84/84, Attention 112/112
4 derivatives checked
- prithivMLmods/Polaris-1.7B-Preview-f32-GGUF: Same weights, converted
- mradermacher/Polaris-1.7B-Preview-GGUF: Same weights, converted
- bunnycore/Qwen3-1.7B-Nyx-V3: Partial fine-tune
- bunnycore/Qwen3-1.7B-Nyx-Fusion: Partial fine-tune