kolibri-1-moe: German pool, Kolibri-1 vocab sweep, recipe and build; eval/neutral-de, eval/longctx, eval/agentic-kolibri

feat(builds): qwen3.8-flash-next-moe, the MoE build for qwen4exp

:hammer: feat(tools): Add abliteration script

Upload recipes/qwen3.8-flash-next.yaml with huggingface_hub

Upload folder using huggingface_hub

Upload recipes/ornith-1.5-9b.yaml with huggingface_hub

Upload folder using huggingface_hub

docs: update README for the pool layout, add RECIPE.md

refactor(tools): move clone.sh into tools and drop the hardcoded path

feat(tools): ship the auto renderer inside the dataset

docs: refactor markdown documentation files

feat(recipes/dsv4): Restore DeepSeek V4 Flash 0731 recipe

refactor(dsv4): Delete/move dsv4 quantization process artefacts

Upload folder using huggingface_hub

Upload recipes/qwen3.8-27b.yaml with huggingface_hub

Upload METHOD.md with huggingface_hub

Upload tools/imcompare.py with huggingface_hub

Upload tools/converge.py with huggingface_hub

Upload tools/build.py with huggingface_hub

Upload recipes/nemotron-3.5-lightning.yaml with huggingface_hub

Upload folder using huggingface_hub

Upload tools/vocab_sweep.py with huggingface_hub

Upload tools/build.py with huggingface_hub

Upload tools/nemotron_fmt.py with huggingface_hub

Upload pool/vocab_sweep/synthetic/nemotron-3.5-lightning.jsonl with huggingface_hub

Upload recipes/nemotron-3.5-lightning.yaml with huggingface_hub

Upload folder using huggingface_hub

README: localise the tokenizer disagreement to non-Latin text

Restructure into a pool + per-model builds; add Muse-Glimmer-30B build

Upload folder using huggingface_hub

Rebuild calibration corpus for DeepSeek-V4-Flash-0731 imatrix

Upload heldout_coding.txt with huggingface_hub

Upload calib_coding.txt with huggingface_hub

initial commit