tiiuae/Falcon-H1-34B-Instruct
Full fine-tune vs tiiuae/Falcon-H1-34B-Base. Fine-tune: 100% of the weights changed, across 72 of 72 layers.
Tensors changed: Embeddings 1/1, MLP 216/216, Norms 217/217, other 504/504, Attention 288/288, lm_head 1/1
10 derivatives checked
- tiiuae/Falcon-H1-34B-Instruct-GGUF: Same weights, converted
- unsloth/Falcon-H1-34B-Instruct-GGUF: Same weights, converted
- tiny-random/falcon-h1: Different architecture
- gabriellarson/Falcon-H1-34B-Instruct-GGUF: Same weights, converted
- mradermacher/Falcon-H1-34B-Instruct-i1-GGUF: Quantization
- yujiepan/falcon-h1-tiny-random: Different architecture
- DevQuasar/tiiuae.Falcon-H1-34B-Instruct-GGUF: Same weights, converted
- mradermacher/Falcon-H1-34B-Instruct-GGUF: Quantization
- tiiuae/Falcon-H1-34B-Instruct-GPTQ-Int4: Quantization
- tiiuae/Falcon-H1-34B-Instruct-GPTQ-Int8: Quantization
1 other repos hold byte-for-byte identical weight files.