Hugging Face's logo

danielharkin21
/
ncm-chat-experimental-generator

next-concept-modeling
ncm
experimental
Model card Files
xet
 Community
ncm-chat-experimental-generator / README.md
danielharkin21's picture
danielharkin21
Upload folder using huggingface_hub
951ac76 verified
preview code
|
Raw History Blame Contribute Delete
1.12 kB
metadata 
license: mit
tags:
  - next-concept-modeling
  - ncm
  - experimental

NCM chat generator (experimental, rough)

Fine-tuned generator weights (fp16, 177 MB) for a small chat demo built on Next-Concept Modeling (Stanford Hazy Research, arXiv 2610.14162, https://github.com/HazyResearch/ncm).

  • Base: the released XSum checkpoint at https://huggingface.co/hazyresearch/ncm-checkpoints (frozen multiscale VQ tokenizer plus prefix-conditioned generator). Only the generator was fine-tuned, for 4000 steps on short public dialogue and instruction data (everyday-conversations, dolly, alpaca; 21.6k examples). Held-out code loss 6.31 to 1.47.
  • Load it with the base repo's config, tokenizer and code (code/ncmchat.py shows how). generator_chat.pt holds a model state dict.
  • Prefix window 128 tokens, reply up to 48 tokens, levels [4,6,10,14,20,26,32,40,48].
  • code/ has the training, ONNX export and parity scripts.
  • Quality is rough: replies are short and often repetitive or wrong. This is a research demo, not an assistant.
  • Browser demo: https://huggingface.co/spaces/danielharkin21/ncm-chat-experimental