Zynerji/Ektome-Qwen3.8-27B-PristinelyUncensored
Upload Ektome-Qwen3.8-27B-Q6_K.gguf with huggingface_hub
Upload README.md with huggingface_hub
Add mtp / speculative-decoding tags for discoverability
Document MTP: how to enable it, measured gains on the 4-bit sibling, and the bf16 caveat
Correct GGUF size (16.8 GB, not ~5 GB) and drop unpublished quants
Remove references to quants that are not published
Remove internal tooling output
Fix YAML front matter: strip CR corruption from repeated edits
Update GGUF status: Q4_K_M rebuilt and verified; other quants unavailable
Rebuilt GGUF: header block_count matches tensors; MTP block present
Document quantized siblings, tensor parity, and excision scope
Add status notice: GGUF withdrawn and being rebuilt (header/tensor mismatch); bf16 unaffected
Use Qwen2VLImageProcessorFast to match base (values already identical)
Restore crc32.txt from Qwen/Qwen3.8-27B (producer does not emit it)
Restore merges.txt from Qwen/Qwen3.8-27B (producer does not emit it)
Restore vocab.json from Qwen/Qwen3.8-27B (producer does not emit it)
Restore video_preprocessor_config.json from Qwen/Qwen3.8-27B (producer does not emit it)
Restore preprocessor_config.json from Qwen/Qwen3.8-27B (producer does not emit it)
Restore LICENSE from Qwen/Qwen3.8-27B (producer does not emit it)
Remove broken GGUF: header declares block_count=65 / nextn_predict_layers=1 but ships only 64 blocks and no MTP tensors
Remove broken GGUF: header declares block_count=65 / nextn_predict_layers=1 but ships only 64 blocks and no MTP tensors
Remove broken GGUF: header declares block_count=65 / nextn_predict_layers=1 but ships only 64 blocks and no MTP tensors
Remove broken GGUF: header declares block_count=65 / nextn_predict_layers=1 but ships only 64 blocks and no MTP tensors
Upload model.safetensors.index.json with huggingface_hub
Upload model-00003-of-00003.safetensors with huggingface_hub
Upload Ektome-Qwen3.8-27B-Q8_0.gguf with huggingface_hub
Upload Ektome-Qwen3.8-27B-Q6_K.gguf with huggingface_hub
Upload Ektome-Qwen3.8-27B-Q5_K_M.gguf with huggingface_hub
Upload Ektome-Qwen3.8-27B-Q4_K_M.gguf with huggingface_hub
Upload ARCH.json with huggingface_hub
Upload ship_Qwen3.8-27B.json with huggingface_hub
Upload README.md with huggingface_hub
Ektome PristinelyUncensored (zero-training weight excision, gen-gated)
initial commit