Skip to content
ML Academy

Books, papers and documentation

Each record says what was actually checked and when. “Verified” means the identity (title, authors, venue/date) was confirmed on the source page; it does not mean every sentence in a lesson is endorsed by the source — claim-level links live on the lesson and insight pages.
  • 811sources
  • 810identity verified
  • 1not verified or restricted
  • 27core reading

Statistics1

  • statsmodels.tools.sm_exceptions (statsmodels source) (opens in a new tab)

    statsmodels developers

    official repoverifiedfree online

    Checked 2026-10-12: Opened 2026-10-12. Source defines PerfectSeparationError (docstring: error due to perfect prediction in discrete models) and a separate PerfectSeparationWarning derived from ModelWarning. Behaviour in a given release (error or warning) was not tested here.

Forecasting and time series1

  • SalesforceAIResearch/gift-eval: GIFT-Eval benchmark and leaderboard (opens in a new tab)

    Salesforce AI Research

    Use for leakage-aware comparison of foundation models (MASE/CRPS ranks, test-data-leakage flag); leaderboard positions change, so cite with date; Day 27.

    official repoverifiedApache-2.0 (code)

    Checked 2026-10-11: GitHub API repo record (Apache-2.0, last push 2026-10-09); README update log (2025-08-05 added TestData Leakage and MASE_Rank columns; 2025-08-25 zero-shot tag; 2025-10-17 replication-code column); leaderboard HF Space huggingface.co/spaces/Salesforce/GIFT-Eval loads (title checked only).

Deep learning5

  • amazon-science/chronos-forecasting: Chronos: Pretrained Models for Time Series Forecasting (opens in a new tab)

    Amazon Science · 2024

    Official code/checkpoints for Chronos, Chronos-Bolt (patch-based, direct multi-step quantile decoder) and Chronos-2; Day 27 lab.

    official repoverifiedApache-2.0 (code)

    Checked 2026-10-11: GitHub API repo record (Apache-2.0, not archived, last push 2026-10-05), releases list (v2.3.2 2026-09-08), README news: Chronos (13 Mar 2024), Chronos-Bolt (26 Nov 2024), Chronos-2 (20 Oct 2025); model table lists amazon/chronos-2 (120M), chronos-bolt tiny/mini/small/base, chronos-t5 sizes.

  • google-research/timesfm: TimesFM (Time Series Foundation Model) (opens in a new tab)

    Google Research · 2024

    Official TimesFM code and version history; check weight licence before production use; Day 27.

    official repoverifiedCode Apache-2.0; TimesFM 3.0 weights under timesfm-non-commercial-license-v1.0 per README (commercial use via Google Cloud services)

    Checked 2026-10-11: GitHub API repo record (Apache-2.0 code, last push 2026-09-29), releases (v3.0.0 2026-08-28, v2.0.2 2026-07-02), README: latest model TimesFM 3.0 (Aug 2026), TimesFM 2.5 released Sept 15 2025, 1.0/2.0 archived; README license notice says TimesFM 3.0 weights use a separate non-commercial license while weights up to 2.5 remain Apache-2.0.

  • Nixtla/nixtla: TimeGPT Python SDK (opens in a new tab)

    Nixtla · Latest release v0.9.0 on 2026-09-21 (GitHub releases)

    Official client SDK for TimeGPT; the repo description (vendor text) names TimeGPT-2.1 as the current model; Day 27.

    official repoverifiedGitHub reports NOASSERTION; model served via Nixtla API

    Checked 2026-10-11: GitHub API repo record (description names 'TimeGPT-2.1'; license reported as NOASSERTION; last push 2026-10-08) and releases list (v0.9.0 2026-09-21).

  • pytorch/examples word_language_model main.py (repackage_hidden) (opens in a new tab)

    PyTorch

    official repoverified

    Checked 2026-10-12: Opened the file. repackage_hidden wraps hidden states in new detached tensors to cut their connection to past computation (recursing into LSTM tuples); the training loop calls it before each batch for non-Transformer models, with a comment that otherwise backpropagation would reach back to the start of the dataset.

  • SalesforceAIResearch/uni2ts: Unified Training of Universal Time Series Forecasting Transformers (opens in a new tab)

    Salesforce AI Research · 2024

    Official Moirai / Moirai-MoE / Moirai 2.0 code; Day 27 lab.

    official repoverifiedApache-2.0 (code)

    Checked 2026-10-11: GitHub API repo record (Apache-2.0, last push 2026-06-02), releases (2.0.0 2025-11-04, 1.2.0 2024-11-28, 1.1.0 2024-06-14), README news: Moirai-2.0-R-small released Aug 2025, Moirai-MoE Oct 2024, Moirai-1.1-R Jun 2024, ICML 2024 oral acceptance May 2024.

NLP and LLMs1

  • rank_bm25: A Collection of BM25 Algorithms in Python (opens in a new tab)

    Dorian Brown (dorianbrown)

    Small Python BM25 implementation for lexical-retrieval labs; supports days 41 and 48.

    official repoverifiedopen source (license not checked)

    Checked 2026-10-11: GitHub repository page loads with title "dorianbrown/rank_bm25 - A Collection of BM25 Algorithms in Python".

MLOps and engineering1

  • wrk2: a constant-throughput, correct-latency-recording variant of wrk (README) (opens in a new tab)

    Gil Tene

    official repoverifiedfree (open-source repository)

    Checked 2026-10-12: README opened 2026-10-12. States that wrk's latency measurement is subject to coordinated omission because each connection waits for a response before the next request, so slow periods are under-represented; wrk2 generates a constant throughput set with -R and measures each latency from when the request should have been sent under that rate. Repository page only; no benchmark numbers used.

Data engineering2

  • dbt-utils (dbt-labs package README) (opens in a new tab)

    dbt Labs

    official repoverifiedfree online (open source)

    Checked 2026-10-12: Opened 2026-10-12. README: unique_combination_of_columns tests that a set of columns is unique together; generate_surrogate_key builds a hashed key and, unlike the older surrogate_key, does not treat nulls and blank strings as identical; expression_is_true checks an expression for every row; recency checks the newest timestamp is within an interval of now; equal_rowcount compares row counts of two relations.

  • pgvector: open-source vector similarity search for Postgres (README) (opens in a new tab)

    Andrew Kane and contributors

    Query-time recall knobs of approximate indexes; defaults can change between versions, re-check.

    official repoverifiedfree

    Checked 2026-10-12: README opened. An approximate index 'trades some recall for speed'; results for queries differ after adding an approximate index; hnsw.ef_search (candidate list size at query time) defaults to 40; ivfflat.probes defaults to 1.