Skill Tracer
Skills › Scientific Agent Skills › hypogenic › sources.md

sources.md

107 lines · skills/hypogenic/references/sources.md at 5f2bc03

Open on GitHub
1# Dated official sources and verification
2
3Review date: **2026-10-01**. The July 2026 review remains historical evidence;
4package, source, dataset revisions, provider contracts and the local tools were
5rechecked for this refresh. Source inspection and mock transport tests are not
6authenticated provider or scientific validation.
7

Dated official sources and verification

Review date: 2026-10-01. The July 2026 review remains historical evidence; package, source, dataset revisions, provider contracts and the local tools were rechecked for this refresh. Source inspection and mock transport tests are not authenticated provider or scientific validation.

Raw lines 1 to 7
8## Release and implementation
9
10- [PyPI package](https://pypi.org/project/hypogenic/) and
11 [official metadata](https://pypi.org/pypi/hypogenic/json): latest release
12 remains 0.3.5, released 2025-07-16, Python >=3.10, beta classifier, MIT,
13 dependencies, artifact hashes and trusted-publisher provenance. The 96,169-byte
14 wheel was downloaded and its SHA-256 matched the metadata.
15- [Official repository](https://github.com/ChicagoHAI/hypothesis-generation),
16 [latest release](https://api.github.com/repos/ChicagoHAI/hypothesis-generation/releases/latest),
17 and [default branch](https://api.github.com/repos/ChicagoHAI/hypothesis-generation/commits/master):
18 v0.3.5 and master `bd37a3129a2f98ee586f545a57b10b59496eedad` remain unchanged.
19- [Pinned release source](https://github.com/ChicagoHAI/hypothesis-generation/tree/8c3800ccae155e333fac5b530afa8abdaac38300)
20 and [pyproject](https://github.com/ChicagoHAI/hypothesis-generation/blob/8c3800ccae155e333fac5b530afa8abdaac38300/pyproject.toml):
21 source/API review targets this exact version rather than an unpinned branch.
22- [Generation CLI](https://github.com/ChicagoHAI/hypothesis-generation/blob/8c3800ccae155e333fac5b530afa8abdaac38300/hypogenic_cmd/generation.py),
23 [inference CLI](https://github.com/ChicagoHAI/hypothesis-generation/blob/8c3800ccae155e333fac5b530afa8abdaac38300/hypogenic_cmd/inference.py),
24 and [logger](https://github.com/ChicagoHAI/hypothesis-generation/blob/8c3800ccae155e333fac5b530afa8abdaac38300/hypogenic/logger_config.py):
25 parser flags and positional arguments, early help exit, swapped logger and
26 hosted-wrapper constructor arguments, small-bank assertion and averaging bug.
27- [Task loader](https://github.com/ChicagoHAI/hypothesis-generation/blob/8c3800ccae155e333fac5b530afa8abdaac38300/hypogenic/tasks.py)
28 and [prompt implementation](https://github.com/ChicagoHAI/hypothesis-generation/blob/8c3800ccae155e333fac5b530afa8abdaac38300/hypogenic/prompt.py):
29 config fields, column JSON, sampling, return order, OOD aliasing, and template
30 substitution. The shipped template rendered locally with synthetic examples.
31- [Model wrappers](https://github.com/ChicagoHAI/hypothesis-generation/tree/8c3800ccae155e333fac5b530afa8abdaac38300/hypogenic/LLM_wrapper):
32 SDK calls, cost table, local registration, model loading, retries, message
33 mutation, block assumptions and error behavior. Executed hosted paths used
34 mock HTTP responses and stubbed heavy import-only dependencies.
35- [Output serializer](https://github.com/ChicagoHAI/hypothesis-generation/blob/8c3800ccae155e333fac5b530afa8abdaac38300/hypogenic/algorithm/update/base.py)
36 and [SummaryInformation](https://github.com/ChicagoHAI/hypothesis-generation/blob/8c3800ccae155e333fac5b530afa8abdaac38300/hypogenic/algorithm/summary_information.py):
37 JSON hypothesis-bank structure and stored statistics.
38- [Redis cache](https://github.com/ChicagoHAI/hypothesis-generation/blob/8c3800ccae155e333fac5b530afa8abdaac38300/hypogenic/LLM_cache.py):
39 cache keys, prompt/response storage and pickle serialization. No Redis server
40 was started or contacted.
41- [PyTorch 2.4.1 release metadata](https://pypi.org/pypi/torch/2.4.1/json):
42 no CPython 3.13 wheels; the upstream package's `torch~=2.4.0` range still
43 prevents the default repository interpreter from installing its full stack.
44- [PyYAML release metadata](https://pypi.org/pypi/PyYAML/json) and
45 [parser documentation](https://pyyaml.org/wiki/PyYAMLDocumentation): reviewed
46 local parser pin advanced to 6.0.3. Valid YAML and rejection of duplicate keys,
47 aliases, anchors, tags, dates and malformed input were exercised locally.

Release and implementation

  • PyPI package and official metadata: latest release remains 0.3.5, released 2025-07-16, Python >=3.10, beta classifier, MIT, dependencies, artifact hashes and trusted-publisher provenance. The 96,169-byte wheel was downloaded and its SHA-256 matched the metadata.
  • Official repository, latest release, and default branch: v0.3.5 and master bd37a3129a2f98ee586f545a57b10b59496eedad remain unchanged.
  • Pinned release source and pyproject: source/API review targets this exact version rather than an unpinned branch.
  • Generation CLI, inference CLI, and logger: parser flags and positional arguments, early help exit, swapped logger and hosted-wrapper constructor arguments, small-bank assertion and averaging bug.
  • Task loader and prompt implementation: config fields, column JSON, sampling, return order, OOD aliasing, and template substitution. The shipped template rendered locally with synthetic examples.
  • Model wrappers: SDK calls, cost table, local registration, model loading, retries, message mutation, block assumptions and error behavior. Executed hosted paths used mock HTTP responses and stubbed heavy import-only dependencies.
  • Output serializer and SummaryInformation: JSON hypothesis-bank structure and stored statistics.
  • Redis cache: cache keys, prompt/response storage and pickle serialization. No Redis server was started or contacted.
  • PyTorch 2.4.1 release metadata: no CPython 3.13 wheels; the upstream package's torch~=2.4.0 range still prevents the default repository interpreter from installing its full stack.
  • PyYAML release metadata and parser documentation: reviewed local parser pin advanced to 6.0.3. Valid YAML and rejection of duplicate keys, aliases, anchors, tags, dates and malformed input were exercised locally.
Raw lines 8 to 48
49## Datasets
51- [HypoBench GitHub repository](https://github.com/ChicagoHAI/HypoBench-datasets)
52 and [default commit](https://api.github.com/repos/ChicagoHAI/HypoBench-datasets/commits/main):
53 remain at `7e4bbc341ee90b7efaa607f67a81543cd68cdf2e`.
54- [Pinned dataset tree](https://github.com/ChicagoHAI/HypoBench-datasets/tree/7e4bbc341ee90b7efaa607f67a81543cd68cdf2e):
55 the deceptive-review config and train/validation/test JSON hashes were
56 rechecked; 800/300/500 rows and three cross-split duplicate groups reproduced.
57 The optional OOD file was acquired only to verify the task's path contract.
58- [Official Hugging Face publication](https://huggingface.co/datasets/ChicagoHAI/HypoGeniC-datasets)
59 and [Hub metadata](https://huggingface.co/api/datasets/ChicagoHAI/HypoGeniC-datasets):
60 revision remains `613860dcbcda9e522a6163ee9edf78c261ebe4bb`.

Datasets

Raw lines 49 to 61
62## Provider contracts and privacy
64- [OpenAI Chat Completions](https://developers.openai.com/api/reference/resources/chat/subresources/completions/methods/create):
65 request and response fields, legacy `max_tokens`, choice/usage semantics and
66 incompatibility with o-series models. The pinned OpenAI 1.40.3 SDK was exercised
67 through HTTPX MockTransport; it sent the documented route and Bearer header.
68- [OpenAI data controls](https://developers.openai.com/api/docs/guides/your-data)
69 and [deprecations](https://developers.openai.com/api/docs/deprecations):
70 endpoint/model-specific storage and availability must be checked at execution
71 time. This review did not query an authenticated model catalog or account.
72- [Anthropic Messages](https://platform.claude.com/docs/en/api/messages/create):
73 top-level system field, content-block response, version header, token parameter,
74 and current temperature restriction. Anthropic 0.32.0 sent the expected route,
75 `x-api-key` and API-version header in mock transport tests.
76- [Anthropic model deprecations](https://platform.claude.com/docs/en/about-claude/model-deprecations):
77 retired model IDs and rejection of non-default temperature on newer models.
78- [Anthropic API retention](https://platform.claude.com/docs/en/manage-claude/api-and-data-retention)
79 and [commercial retention](https://privacy.claude.com/en/articles/7996866-how-long-do-you-store-my-organization-s-data):
80 standard deletion policy, exceptions, ZDR eligibility, and covered-model
81 30-day requirements. No account-specific arrangement was verified.
82- [OpenAI 1.40.3 client source](https://github.com/openai/openai-python/blob/v1.40.3/src/openai/_client.py)
83 and [Anthropic 0.32.0 client source](https://github.com/anthropics/anthropic-sdk-python/blob/v0.32.0/src/anthropic/_client.py):
84 automatic credential lookup and base-URL environment overrides. These were
85 also inspected in the installed pinned SDKs; no credential values were read.

Provider contracts and privacy

  • OpenAI Chat Completions: request and response fields, legacy max_tokens, choice/usage semantics and incompatibility with o-series models. The pinned OpenAI 1.40.3 SDK was exercised through HTTPX MockTransport; it sent the documented route and Bearer header.
  • OpenAI data controls and deprecations: endpoint/model-specific storage and availability must be checked at execution time. This review did not query an authenticated model catalog or account.
  • Anthropic Messages: top-level system field, content-block response, version header, token parameter, and current temperature restriction. Anthropic 0.32.0 sent the expected route, x-api-key and API-version header in mock transport tests.
  • Anthropic model deprecations: retired model IDs and rejection of non-default temperature on newer models.
  • Anthropic API retention and commercial retention: standard deletion policy, exceptions, ZDR eligibility, and covered-model 30-day requirements. No account-specific arrangement was verified.
  • OpenAI 1.40.3 client source and Anthropic 0.32.0 client source: automatic credential lookup and base-URL environment overrides. These were also inspected in the installed pinned SDKs; no credential values were read.
Raw lines 62 to 86
87## Local wrappers and scientific scope
89- [Transformers 4.45.2 pipelines](https://huggingface.co/docs/transformers/v4.45.2/en/main_classes/pipelines)
90 and [installation/offline mode](https://huggingface.co/docs/transformers/v4.45.2/en/installation):
91 model/path loading and offline configuration compatible with the pinned
92 4.45.x line. No model weights or local inference were tested.
93- [vLLM 0.6.2 LLM API](https://docs.vllm.ai/en/v0.6.2/dev/offline_inference/llm.html):
94 offline model, sampling and generation interfaces; reviewed alongside the
95 package wrapper, not validated on GPU hardware.
96- [Hypothesis Generation with Large Language Models](https://aclanthology.org/2024.nlp4science-1.10/):
97 Zhou et al., NLP4Science 2024, DOI `10.18653/v1/2024.nlp4science-1.10`;
98 data-driven candidate generation and classification evaluations.
99- [Literature Meets Data](https://arxiv.org/abs/2410.17309): Liu et al.,
100 version 3 dated 2025-01-08; HypoRefine and literature/data integration.
101- [HypoBench](https://arxiv.org/abs/2504.11524): Liu et al., version 2
102 dated 2026-02-10; seven real-world tasks, five synthetic families and 194
103 datasets. Publication metadata does not establish scientific validity of an
104 individual generated hypothesis.
105- [HypoBench OpenReview record](https://openreview.net/forum?id=cizEoSePyT):
106 current retrieval reached a browser challenge. No current review/acceptance
107 status is asserted from that response.

Local wrappers and scientific scope

  • Transformers 4.45.2 pipelines and installation/offline mode: model/path loading and offline configuration compatible with the pinned 4.45.x line. No model weights or local inference were tested.
  • vLLM 0.6.2 LLM API: offline model, sampling and generation interfaces; reviewed alongside the package wrapper, not validated on GPU hardware.
  • Hypothesis Generation with Large Language Models: Zhou et al., NLP4Science 2024, DOI 10.18653/v1/2024.nlp4science-1.10; data-driven candidate generation and classification evaluations.
  • Literature Meets Data: Liu et al., version 3 dated 2025-01-08; HypoRefine and literature/data integration.
  • HypoBench: Liu et al., version 2 dated 2026-02-10; seven real-world tasks, five synthetic families and 194 datasets. Publication metadata does not establish scientific validity of an individual generated hypothesis.
  • HypoBench OpenReview record: current retrieval reached a browser challenge. No current review/acceptance status is asserted from that response.
Raw lines 87 to 107