adidsh/indic-speak-int8-onnx
Full fine-tune vs meta-llama/Llama-3.2-3B. adidsh/indic-speak-int8-onnx is a GGUF (Q8_0) of different weights. Fine-tune: 100% of the weights changed, across 28 of 28 layers.
Tensors changed: Norms 57/57, MLP 84/84, Attention 112/112