prithivMLmods/Blitzar-Coder-4B-F.1
Full fine-tune vs prithivMLmods/Qwen3-4B-ft-bf16. Fine-tune: 90% of the weights changed, across 36 of 36 layers.
Tensors changed: Embeddings 0/1, Norms 81/145, MLP 108/108, Attention 144/144
5 derivatives checked
- mradermacher/Blitzar-Coder-4B-F.1-i1-GGUF: Quantization
- prithivMLmods/Blitzar-Coder-4B-F.1-GGUF: Same weights, converted
- mradermacher/Blitzar-Coder-4B-F.1-GGUF: Same weights, converted
- DavidAU/Qwen3-Blitzar-Coder-F1-6B-Brainstorm20x: Partial fine-tune (fewer layers)
- sillykiwi/Qwen3-4B-PastelCoder: Full fine-tune