arcee-ai/SuperNova-Medius
Full fine-tune vs Qwen/Qwen2.5-14B. Fine-tune: 100% of the weights changed, across 48 of 48 layers.
Tensors changed: Embeddings 1/1, Norms 97/97, MLP 144/144, Attention 336/336, lm_head 1/1
10 derivatives checked
- bartowski/SuperNova-Medius-GGUF: Same weights, converted
- arcee-ai/SuperNova-Medius-GGUF: Same weights, converted
- mradermacher/SuperNova-Medius-i1-GGUF: Quantization
- mradermacher/SuperNova-Medius-GGUF: Quantization
- allura-org/TQ2.5-14B-Neon-v1: Partial fine-tune
- allura-org/TQ2.5-14B-Sugarquill-v1: LoRA-style
- nlpguy/Lion-Lamarck-v.1.0.9: Full fine-tune
- nlpguy/Lion-Lamarck-v.1.0.8: Full fine-tune
- Sakalti/Saba2-14B-Preview: Full fine-tune
- win10/EVA-Meissa-mini-pro: Full fine-tune