← The Independent 100: a study

Cody Blakeney

#61 in The Basement Labs, the curator’s rank and section @code_star Relevance to our work: 3 of 5

Bracketed tags are evidence labels: [Verified] means we read the page, file or API response ourselves, [Reported] means a source says so and we did not check the underlying thing, and [Inference] is our own reasoning.

Credit

The Independent 100 list was curated by dan (@irl_danB) and published on 2026-09-24 at independent.prose.md (announcement). The ranks and section names are his, and number 100 is his open wild-card slot. This study is ours and is not endorsed by him or by anyone on the list.

What they publicly do

[Reported] Leads research at Arcee AI; formerly Data Research Lead at Databricks Mosaic (X profile x.com/code_star). [Verified] Announced Arcee Trinity Large as 'the return of American OSS' and describes large synthetic-data and multilingual curation work for it (x.com/code_star/status/2016279734985097532). Writes the Substack tokenlibrary.ai.

Public work

  • Arcee Trinity Large (400B sparse MoE) model OpenMDW-1.1 last activity 2026-05-28

    Open-weight MoE trained on 17T tokens in three phases. The technical report is arXiv 2602.17004. Also Thinking, Mini and Nano variants with GGUF, FP8 and NVFP4 builds.

  • Arcee Trinity Large Technical Report paper no license stated last activity 2026-02-19

    Data mixing, midtraining phases, synthetic code generation (800B+ tokens).

  • llm-hparam-flop-calculator repo Apache-2.0 last activity 2025-11-19 · 9 GitHub stars on 2026-09-24

    Browser tool for FLOP and hyperparameter back-of-envelope sizing.

  • Acceptable Hyperparameters and How to Find Them blog no license stated last activity 2025-11-15

    Companion post to the FLOP calculator.

  • composer-hydra-template repo Apache-2.0 last activity 2022-11-15 · 18 GitHub stars on 2026-09-24

    Template for MosaicML Composer training with Hydra configs.

Licenses of GitHub repositories are from the authenticated GitHub license API on 2026-09-25, one row per repository in evidence/licenses.jsonl (“no license file” means the API found none; “license file present, not recognised by GitHub” means GitHub could not classify it). Other licenses are as the source states. Check the license file before reusing anything.

Relevance to our work

3 of 5 lenses: training, local-models, security

[Verified] Trinity's open weights under OpenMDW-1.1 (a permissive open-model license) and the GGUF builds are candidates for local serving, and the technical report documents data-mixture methods. [Verified] His 2026-09-18 posts on plaintext secrets in scraped code, and on hijacked agents with Google, Slack and GitHub access as the real risk, bear on prompt-injection and secret-leak risk in agent tools (x.com/code_star/status/2101009572693778512, x.com/code_star/status/2100847565302538722). Little agent tooling of his own.

Scored 0 to 5 against the study’s relevance rubric: 5 means public, openly licensed work squarely in our lanes that we could study or adopt now, and 1 means little connection to our work. The score is one reviewer’s judgement of how much this public work bears on our own work. It is not a ranking of people and says nothing about the person or the quality of their work.

What we would study next

  1. Trinity technical report, data-mixture and synthetic-code sections (arxiv.org/abs/2602.17004). A method reference for fine-tuning a local model on agent transcripts. It also informs which open models to trial behind a coding harness. Effort: medium.
  2. Try a Trinity Nano or Mini GGUF as a cheap local sub-agent model (huggingface.co/arcee-ai). OpenMDW-1.1 is permissive, so a local-model benchmark is low risk. Effort: small.

Before reusing anything: The Trinity HF model cards list 'license: other, license_name: openmdw-1.1' (OpenMDW License 1.1, a permissive model-weights license; not an OSI software license). Check the LICENSE file in each repo for attribution terms. His GitHub tools are Apache-2.0. Do not attribute mergekit to him: it is an Arcee org repo (LGPL-3.0) with other authors.

Sources

Overall confidence in this record: Medium. Sources were read on 2026-09-24 and every link was checked on 2026-09-25. The full record is line 61 of people.jsonl.

Limits

Not verified: His specific authorship role on the Trinity technical report (the author list was not read in full). What Arcee has built in the last ~6 months (teased 2026-09-16, not yet public). The full OpenMDW-1.1 text.

This record comes from a first-pass survey of public material by an AI agent session on 2026-09-24, edited for publication on 2026-09-25. Nothing was installed or run for it. Roles, stars and licenses change; see the study’s limits.

Request a correction

Is something about this record wrong or out of date, or would you rather not be covered? Open a correction (a GitHub issue form with this page’s address filled in). A request to remove a record is honoured without argument; accepted corrections are recorded in the changelog with the date.

Help fix it

Each opens a GitHub issue form with this page’s address filled in. You need a GitHub account.

Made by Joshua Nowak (ZestStream) with AI coding agents. Not affiliated with or funded by Jeffrey Emanuel; Joshua uses his public tools and is a paying subscriber to jeffreys-skills.md.

Source on GitHub MIT licensed