mradermacher/Tashkeel-350M-v2-GGUF
Same weights, converted vs Etherll/Tashkeel-350M-v2. Same weights stored in another dtype (BF16 → F16/F32), plus 124 new tensors.
Tensors changed: Embeddings 0/1, Norms 0/65, Attention 0/16
Same weights, converted vs Etherll/Tashkeel-350M-v2. Same weights stored in another dtype (BF16 → F16/F32), plus 124 new tensors.
Tensors changed: Embeddings 0/1, Norms 0/65, Attention 0/16