early-failure-detection-cot
Research on LLM reasoning choke points, failure onset & counterfactual commitment in process supervision (ProcessBench, Llama-3.1, Qwen2.5)
What XERJ indexed
Snapshot
The whole repository at commit 7da2240a72 (2026-10-27). main is 9 commits ahead.
Selected work
“ChokeBench predictive validity (held-out halves and models); DRA test…”. 16 files, 2,218 lines in results, scripts, paper.
Licence
The repository
LICENSE file checked at the pinned commit; the author agreed to inclusion.
The XERJ corpus
The author agreed by email on 2026-10-28. Same terms as the repository. Answers name the repo, the commit and the file:line.
selected for the XERJ Corpus Hublicence clearedsuggested use: adapt with attribution
chain-of-thoughtllmreasoningresearchai-safetyinterpretabilityllm-reasoningprocess-supervision
Example questions
- How does early-failure-detection-cot implement chain-of-thought?
- What does results do in early-failure-detection-cot?
- Show the llm code in early-failure-detection-cot.
Inside the indexed corpus
- 541 files, 65.0 MB in total, about 123.0 KB per file.
- The selected commit changed 16 of those files (3.0%).
- Languages by size: Jupyter Notebook 47.5%, HTML 25.5%, Python 19.9%, TeX 6.3%, Shell 0.6%.
- On GitHub since 2026-09-15, last push 2026-10-28, 104 commits on main.
- The pin is 9 commits behind main; the corpus answers from the pinned code, not the newest.
Licence on GitHub and in the corpus
GitHub reported MIT for HarshDubey23/early-failure-detection-cot on 2026-10-31, and the LICENSE file at commit 7da2240a72 was reviewed as MIT. They match, so answers from this corpus carry the same terms as the live repository.
Contributors
Harsh Dubey (@HarshDubey23) has 1 selected repository on the hub.
Using XERJ with early-failure-detection-cot
- Search Jupyter notebooks (.ipynb) and the .py beside them Jupyter Notebook is 47.5% of the code by size.
- RAG without a vector database: one hybrid index The hub files it under AI, prompts and agents.
- ripgrep vs a code index for coding agents: 30 tasks
- Best MCP servers for local code search in 2026, compared
The prompt
Paste it into your AI agent. It clones HarshDubey23/early-failure-detection-cot at the pinned commit and indexes it on your machine.
Search the GitHub repository HarshDubey23/early-failure-detection-cot with XERJ (https://hub.xerj.org/corpus/harshdubey23--early-failure-detection-cot). Run these in a shell, in order, and check each result before the next: 1. xerj --help (success: it prints usage). If xerj is missing: curl -fsSL https://xerj.org/get | sh 2. git clone https://github.com/HarshDubey23/early-failure-detection-cot early-failure-detection-cot && git -C early-failure-detection-cot checkout 7da2240a724fbacf5cc3388f8f6e5bd028cf2940 3. Start a local node and leave it running: xerj --insecure --data-dir ./xerj-data 4. xerj autoindex ./early-failure-detection-cot --url http://127.0.0.1:9200 --prefix early-failure-detection --progress plain 5. xerj autoindex map --url http://127.0.0.1:9200 Then answer my question from the indexed files and cite the ax_path of every hit. Query syntax: https://xerj.org/llms.txt Licence: MIT. Keep the copyright notice and credit the authors.
Related
Harsh DubeyAI, prompts and agents corporabtrblocksflexragaruintelligence/aml-coreSiddarthb07/AnimaStartLuxLabs/StartLux-Decision
Stats: GitHub API, fetched 2026-10-31. Snapshot: the git tree at the pinned commit. Selected work and consent: the XERJ registry.