Badtheorylabs/interference-search
Reason over explicit states, many branches at once: merge duplicates, cancel dead ends, advance together.
About the project
A search method that reasons over explicit states instead of a language model's linear transcript: many branches expand at once, duplicates merge, dead ends are dropped by a trained judge, and survivors advance together. The repo ships the code, the trained judge, raw results and a research log.
Useful for
- Run Countdown search on CPU and compare step counts against Qwen3-1.7B reasoning in text
- Train the judge on four- and five-number problems and test it on seven-number ones
- Reproduce the numbers in paper/CLAIMS.md using the listed files and commands
README summarized by DeepSeek V4.1 Flash. Details may be inaccurate.
Why it’s trending
- 92 stars today.
- The repository is 0 days old and already has 92 stars. With less than two weeks of history, there's no usual pace to compare the spike against yet.
- Top new repositories this week: #188.
Stars per day
Bars are daily stars, the line is the usual pace. Red marks spike days.
Numbers
- Total stars
- 92
- Today
- 92
- Forks
- 3
- Issues and pull requests
- 0
- Watchers
- 0
- Language
- Python
- License
- Apache-2.0
- Created
- September 25, 2026
- Last push
- September 25, 2026
Star trust
Growth looks organic: forks and discussion are in line with active projects, and stars arrive unevenly, the way people give them.
These are heuristics, not a verdict: we judge by the repository’s behavior, not by a list of stargazers.
Spotted in
- September 25, 2026Top new repositories this week: #188
Share
README badge
Paste it into your README — the badge shows the current magnitude and links to this page.
[](https://gitnova.dev/en/r/Badtheorylabs/interference-search)More in this category
-
6.3
ollaya-dev/ollaya
A local runtime for decision models: pulls and serves classification and routing models behind a TypeSafe-compatible API. Like Ollama, but for models that return probabilities instead of text.
-
6.2
jaredpalmer/kev
kev is a LoRA adapter with a small readout head on top of Qwen2.5-0.5B that answers many typed questions about a document in a single forward pass, returning calibrated probabilities instead of text.
-
6.0
NVIDIA/Model-Optimizer
NVIDIA library for compressing and accelerating models: quantization, pruning, distillation, NAS and speculative decoding with export to TensorRT-LLM, vLLM, SGLang. For ML engineers preparing models for deployment.
-
5.5
togethercomputer/tev1
Open dataset and fine-tuning recipe for Qwen3.5-4B that picks one of 2–24 options given context and a question. The repo ships data, build scripts, Together AI training and evaluation tooling.
-
5.4
nokia-applied-research/AnyJev
Turns any LLM into a decision model with typed questions (choice, yes/no, score) that return real probabilities read from the next-token distribution, without training. L0 removes position and label-prior bias; L1 adds…
-
5.3
deepopen-com/deepopen
Open-source non-autoregressive System 1 decision engine for typed decisions: returns typed answers (choice, score, noul) in a single forward pass across 100+ languages without generating text. Suited for…