Earlier sources: Artificial Intelligence

Earlier sources

Up to ten sources with readable evidence, ordered by their feed timestamp. These archive sources can be older than the live fourteen-day window. This list is a sample, not a measure of total activity. Dates distinguish paper publication, repository creation, model/dataset activity, Hacker News posts, and KEV catalog additions. “First archived” records when SYMTRI first stored the signal.

1. Hugging Face · Model / dataset date 2026-10-16 · First archived 2026-10-16

   MarcelEsmond/Annoy-PyEdu-Rs-Raw — dataset release ↗

   # Annoy: This should be a paper Title 📑 Paper &nbsp&nbsp | &nbsp&nbsp 🌐 Project Page &nbsp&nbsp | &nbsp&nbsp 💾 Released Resources &nbsp&nbsp | &nbsp&nbsp 📦 Repo We release the raw data for our processed PythonEdu-Rs dataset, adopted from the original dataset from HuggingFaceTB team. The data format for each line in the `0_368500_filtered_v2_ds25.sced.jsonl` is as follows: ``` { "problem_description": , "io_requirements": , "refcode": , "funcname": , "ios": [ { "input": , "output": }, ... ], "source": , "category": , "meta": } ``` Some of the `ios` are empty. The reason is that when executi

2. Hugging Face · Model / dataset date 2026-10-16 · First archived 2026-10-07

   punctualprocrastinator/relational-generator-runs — dataset update ↗

   # relational-generator-runs Synthetic relational-database corpora, Relational Transformer checkpoints pretrained on them, and evaluation tables for the generator study. Everything here is generated data or derived from public RelBench; no private records.

3. Hugging Face · Model / dataset date 2026-10-16 · First archived 2026-10-16

   lamsheeper-data-attribution/vtok101-attribution-baselines — dataset update ↗

   # vtok101 attribution scores (Function Bindings) The attribution scores of "Elucidating the Design Space of LM Data Attribution" on **Function Bindings**, for the LoRA adapters in [lamsheeper-data-attribution/Qwen3.5-4B-d0-vtok101-lora-seeds](https://huggingface.co/lamsheeper-data-attribution/Qwen3.5-4B-d0-vtok101-lora-seeds). The CATT code release reads them with `catt fetch results --task fb` and rebuilds the paper's figures and tables from them with `catt report results`. ## Layout {run}/{group}/scores.{method}.{arm}.npz one [query, train] score matrix {run}/{group}/scores.npz the group's f

4. Hugging Face · Model / dataset date 2026-10-16 · First archived 2026-10-16

   Sologid/N-Atlas-cassava-Umuahia — model update ↗

   # N-ATLaS Cassava Advisor: Umuahia Cassava can lose value at every handoff: when roots are bruised at harvest, left waiting, moved on rough roads, or dried too slowly. I’m building this advisor to give people in Abia practical guidance on those weak points—and, over time, to support better coordination from farm gate to processing. This is an early prototype. The model gives cassava post-harvest advice; it does **not** arrange pickups, manage transport, aggregate roots, or buy or sell cassava. Under the hood, it is a PEFT QLoRA adapter for the gated 8B `NCAIR1/N-ATLaS` model, not a standalone

5. Hacker News · Posted on Hacker News 2026-10-16 · First archived 2026-10-16

   How the Claude Code team does code review ↗

   Hacker News discussion.

6. Hacker News · Posted on Hacker News 2026-10-16 · First archived 2026-10-16

   AlbertCode, an AI software engineering agent for building applications ↗

   Hacker News discussion.

7. Hugging Face · Model / dataset date 2026-10-16 · First archived 2026-10-16

   litert-community/NVIDIA-Nemotron-Nano-9B-v2 — model release ↗

   # NVIDIA-Nemotron-Nano-9B-v2 — LiteRT-LM [nvidia/NVIDIA-Nemotron-Nano-9B-v2](https://huggingface.co/nvidia/NVIDIA-Nemotron-Nano-9B-v2) converted to the **LiteRT-LM** (`.litertlm`) format for Google's [LiteRT-LM](https://github.com/google-ai-edge/LiteRT-LM) runtime: an int8 file and an int4 file, built on the Nemotron-H conversion path of [hf-to-litertlm](https://github.com/john-rocky/hf-to-litertlm) (`nemotron_h_work/`). Converted with litert-torch (commit 115a136 with the Nemotron-H patch) and tested on litert-lm 0.18.0 on a Mac Studio M4 Max, where the int4 file decodes at 45.5 tok/s on the

8. Hugging Face · Model / dataset date 2026-10-16 · First archived 2026-10-16

   openeurollm/datamix-0.7b-openthoughts_3-100bt — model update ↗

   # Model Details This is a decoder-only model with approximately 0.7B parameters. The architecture largely follows the Qwen-3 design, with the following key hyperparameters: - Hidden Size: 1024 - Attention Heads: 16 - Layers: 28 - Sequence Length: 4096 # Training Data The total token budget for training is 100 billions tokens. The training mixture is comprised of Nemotron-CC high-actual (85%) and OpenThoughts3-1.2M dataset (15%). # Tokenizer The model utilizes custom [openeurollm tokenizer](https://huggingface.co/openeurollm/tokenizer-256k) with a 262K vocabulary size. # Training Information Th

9. Hacker News · Posted on Hacker News 2026-10-16 · First archived 2026-10-16

   Show HN: A music video made end to end by Claude Code, full production repo ↗

   Hacker News discussion.

10. GitHub · Repository created 2026-10-16 · First archived 2026-10-16

    notaaaronfr/myelinate ↗

    Self-improving skills and memory for any AI agent framework, with governance built in

How to assess evidence about Artificial Intelligence

Check each original source for its methods, release status, limitations, and supporting data. A repository announcement, a preprint, and a published study provide different kinds of evidence. SYMTRI’s subject classification helps discovery; it does not independently validate the source’s claims.

See the collection and counting methodology ↗ · Read the weekly synthesis ↗

Follow a thread

Explore related regions

These navigation suggestions are part of the map’s editorial structure, rather than measured evidence of a connection.