No comments yet.Be the first to comment.
DeepSeek-V4-Flash-0731-Latent-Reasoning. A self-contained model that does thinking in latent space, NVFP4-quantized, with a production vllm form for serving runtime.
https://huggingface.co/nmitchko/DeepSeek-V4-Flash-0731-Latent-Reasoning
Last time I grafted aCoLaR head(Compressed Latent Reasoning) onto DeepSeek Flash v4. The work had all the makings of a demo. A head that lets the model think in latent space. A learned stop head that decides when it has thought enough. A riddle that showed why plain autoregressive generation regurgitates cached answers instead of reasoning.
There was one honest problem though: the head was anadapter. Something bolted on the side. To serve it you had to assemble the base model, the head, the stop criterion, and a custom runtime by hand, then hope the pieces fit. The weights lived in one place. The inference machinery in another. “Here’s how you run it” was a jump-through-hoops story.
This edition closes that gap. The work is now acomplete, self-contained model. Every weight needed to serve it ships in one repository. The backbone is quantized down to NVFP4 so it fits on real silicon. And the latent loop is driven by a proper, benchmarked serving runtime.
Model:nmitchko/DeepSeek-V4-Flash-0731-Latent-Reasoning
The whole point of this edition is packaging. The old release was a head you had to attach. The new one is a model.
Every weight needed to serve it now sits in one HuggingFace repo:
Because the weights are complete, the model card can finally report areal benchmarkinstead of “more benchmarks soon.”
BBH (BIG-Bench Hard),cot_zeroshot, 27 subtasks: aggregate 0.94 ± 0.008.Measured with lm-evaluation-harness 0.4.12 against an OpenAI-compatible endpoint. Deepseek latent reevaluated the results after giving separate 0.94 vs 0.88 exact match. Thinking enabled. 50 items per subtask, 1350 items total.
The pattern is exactly what you would hope for from a latent reasoning model. It is strongest where reasoning means multi-step state tracking.tracking_shuffled_objects,boolean_expressions,formal_fallacies,penguins_in_a_tableall hit 1.00. It is weakest on the mechanical, syntax-heavy jobs.dyck_languages(bracket matching) sits at 0.26, the clear outlier. That is a genuine weakness, not a measurement artifact.