mradermacher/SciThinker-30B-GGUF
Full fine-tune vs OpenMOSS-Team/SciThinker-30B. mradermacher/SciThinker-30B-GGUF is a GGUF (Q8_0) of different weights. Fine-tune: 100% of the weights changed, across 48 of 48 layers.
Tensors changed: Embeddings 1/1, Norms 71/193, MLP 48/48, Attention 192/192, lm_head 1/1