athirdpath/Theia-21B-v1_pretrained
Full fine-tune vs TheDrummer/Theia-21B-v1. Fine-tune: 100% of the weights changed, across 70 of 70 layers.
Tensors changed: Embeddings 1/1, Norms 0/141, MLP 210/210, Attention 280/280, lm_head 1/1
3 derivatives checked
- mradermacher/Theia-21B-v1_pretrained-i1-GGUF: Quantization
- mradermacher/Theia-21B-v1_pretrained-GGUF: Quantization
- Raven-Pro/mixup1001: Partial fine-tune