ubergarm/DeepSeek-V3.1-GGUF
Upload folder using huggingface_hub
Add amazing new smol-IQ4_KSS
Add bf16 baseline PPL
Upload folder using huggingface_hub
Add IQ2_KT
Upload folder using huggingface_hub
Show "pure" Q4_0 for comparison
Upload folder using huggingface_hub
Add IQ3_K
Upload folder using huggingface_hub
Uploading IQ4_KS and IQ4_K
Okay the initial set is done
Add IQ2_KS and more perplexity graph data
Upload folder using huggingface_hub
Upload folder using huggingface_hub
Add IQ1_KT
Upload folder using huggingface_hub
add perplexity graph and upload more
Upload folder using huggingface_hub
Upload folder using huggingface_hub
Uploading IQ2_KL and preparing IQ3_KS
Add baseline Q8_0 perplexity
Preparing IQ2_KL 231.206 GiB (2.960 BPW)
Upload folder using huggingface_hub
Add IQ1_S and some perplexity measurements.
Upload imatrix-DeepSeek-V3.1-Q8_0.dat with huggingface_hub
Add note to use `--chat-template deepseek3` for now
Upload folder using huggingface_hub
Uploading iq5_k
Upload imatrix-DeepSeek-V3.1-Q8_0.dat with huggingface_hub
initial commit
initial commit