mradermacher/ABot-OCR-GGUF
Full fine-tune (fewer layers) vs acvlab/ABot-OCR. mradermacher/ABot-OCR-GGUF matches part of acvlab/ABot-OCR (310 of its 625 tensors; the rest of acvlab/ABot-OCR has no counterpart). On the shared part: mradermacher/ABot-OCR-GGUF is a GGUF (F16) of different weights. Fine-tune: 100% of the weights changed, across 28 of 28 layers.
Tensors changed: Embeddings 1/1, Norms 108/113, MLP 84/84, Attention 112/112