Presentation: goal wording, layout and programme status

README.md

@@ -1,8 +1,8 @@
 ---
 title: Optitransfer
 emoji: 🔬
-colorFrom: indigo
-colorTo: blue
+colorFrom: gray
+colorTo: gray
 sdk: static
 pinned: true
 tags:
@@ -15,106 +15,189 @@ tags:
   - reproducible-evaluation
   - benchmarks
   - leaderboards
+  - crdt
 ---
 
 <div align="center">
 
 # Optitransfer
 
-**Converge: building frontier-level open models from the work the open-source community has already published.**
+**Open models built to place at the top of their class on public leaderboards.**
 
-*One repeatable process at every model class. Every result is measured against its base model and published with per-item evidence.*
+*Built from fine-tunes the open-source community has already published. Measured against the base model. Published with the evidence.*
 
-[Models](https://huggingface.co/Optitransfer) · [crdt-merge](https://github.com/mgillr/crdt-merge) · [acfa-rs](https://github.com/mgillr/acfa-rs) · [Paper](https://arxiv.org/abs/2605.19373)
+---
+
+[![Models](https://img.shields.io/badge/models-converge%20v1%20%C2%B7%20v2-blue)](https://huggingface.co/Optitransfer)
+[![Base](https://img.shields.io/badge/base-Qwen2.5--7B--Instruct-blueviolet)](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct)
+[![Paper](https://img.shields.io/badge/arXiv-2605.19373-b31b1b)](https://arxiv.org/abs/2605.19373)
+[![Paper](https://img.shields.io/badge/arXiv-2607.10305-b31b1b)](https://arxiv.org/abs/2607.10305)
+[![crdt-merge](https://img.shields.io/pypi/v/crdt-merge?label=crdt-merge&color=blue)](https://pypi.org/project/crdt-merge/)
+[![Downloads](https://img.shields.io/pypi/dm/crdt-merge?label=downloads&color=brightgreen)](https://pepy.tech/projects/crdt-merge)
 
 </div>
 
 ---
 
-## Purpose
+## Thesis
+
+Thousands of fine-tuned models sit on public repositories. Each adds a skill to a shared base model. Their training is already paid for.
+
+**Hypothesis:** that capability can be consolidated into one model per class, without pre-training and without trading one skill for another.
+
+**Converge** is the programme that tests it, one release at a time, in public.
+
+---
+
+## Goal
+
+**Top placements in class, across the public leaderboards.**
+
+| | |
+|:--|:--|
+| **Scope** | 7B first. Then each larger size class and other model families. |
+| **Bar** | Improve on reasoning, mathematics, code, instruction following and knowledge, with no statistically significant regression against the base model. |
+| **Done** | A class is complete when no public fine-tune improves any axis further. |
+
+---
+
+## Where We Are
+
+Measured results, stated as they are.
+
+| | |
+|:--|:--|
+| **v2 against its base** | 12 benchmarks: 2 improved, 2 regressed, 8 with no measurable difference |
+| **Improved** | GSM8K under HELM's protocol: 87.0 against 83.3 (+3.7, calibrated). MMLU under HELM's protocol: +0.47 (calibrated). |
+| **Regressed** | MATH Level 5: -5.74. MMLU-Pro: -1.89. |
+| **Defect found** | A defect in the merge that produced v1, inherited by v2. Corrected. The corrected v1 measures at its base model's level on MMLU-Pro (TIGER-Lab). |
+| **Not yet shown** | A release that improves on its base with no regression. That is the test the next release must pass. |
+
+Full tables, confidence intervals and limitations are on each model card.
+
+---
+
+## Why It Matters
+
+| | |
+|:--|:--|
+| **Science** | Combining fine-tuned models usually trades one capability for another, in our first releases as well. Whether consolidation can hold on every axis at once is an open, testable question. |
+| **Economics** | Consolidation reuses training the community has already paid for. If the hypothesis holds, the marginal cost of a stronger open model is evaluation compute, not pre-training. |
+| **Adoption** | Organisations that keep AI inside their own boundary need capability without dependence on one vendor, and a record of how each model change was made. |
+
+---
+
+## Process
+
+One process, run the same way at every class.
+
+| | Stage | What happens |
+|:--|:--|:--|
+| 1 | **Start** | A strong open base model in the class |
+| 2 | **Discover** | The compatible fine-tunes the community has published for it |
+| 3 | **Assimilate** | They are brought into one model |
+| 4 | **Measure** | Every axis, against the base model, on identical items and public protocols |
+| 5 | **Release** | Only when the model holds against its base and the previous release on the whole board |
+| 6 | **Repeat** | Until nothing more improves, then the next class |
+
+The construction method is proprietary. The evaluation is public. The release rule in step 5 applies from the corrected releases on; v1 and v2 predate it, and their cards list their regressions.
+
+---
+
+## Models
+
+Releases are published under this organisation from converge v3. The research behind them is on the founder's page, [@Optitransfer](https://huggingface.co/Optitransfer).
 
-Frontier AI is concentrated in a handful of laboratories, because each generation of pre-training costs more than
-almost anyone else can spend. The open ecosystem has already paid that cost many times over. Thousands of fine-tuned
-models sit on public repositories, each one somebody's GPU-hours and each adding a skill to a shared base: mathematics,
-code, instruction following, a domain, a language.
+### Releases
+
+| Model | Status |
+|:--|:--|
+| converge v3 | Next. Released when it passes the release rule. |
+| converge v4 | Planned. Same rule. |
+| Larger classes | Planned. Same process. |
+
+### Research
+
+| Model | Status |
+|:--|:--|
+| [converge v1](https://huggingface.co/Optitransfer/Qwen2.5-7B-Instruct-converge-collective-v1) | Public. First 7B experiment, on Qwen2.5-7B-Instruct. Carries the defect; see the notice on its card. |
+| [converge v2](https://huggingface.co/Optitransfer/Qwen2.5-7B-Instruct-converge-collective-v2) | Public. Built on v1, with the 12-benchmark board above. Inherits the defect. |
+| converge v1, corrected | Built. Full board in progress. |
+| converge v2, corrected | Built. Full board in progress. |
+
+The corrected models' boards are published when complete, whichever way they fall. Release follows the release rule.
+
+---
 
-**Converge** is built to turn that work into frontier-level open models. It does not out-train the laboratories. It brings what
-the community has built into a single model for each model class, measures every step, keeps every gain and
-publishes every regression. The model is never finished: it improves release after release.
+## Evidence Standard
 
-## The goal
+| Principle | Meaning |
+|:--|:--|
+| **Paired** | Every score is the model minus its base on the same items, with a 95% confidence interval and an exact significance test |
+| **Regressions reported** | A card lists what went down as prominently as what went up |
+| **Calibration labelled** | Calibrated means the harness first reproduced the base model's published score. Every result is marked either way |
+| **Public protocols** | HELM-protocol GSM8K and MMLU, MMLU-Pro (TIGER-Lab), ZeroEval, EvalPlus HumanEval+ and MBPP+, MATH, AIME, ARC, IFEval |
+| **Corrected in public** | When a defect is found, the affected cards say so first |
 
-**Reach frontier-level performance in every model class.** The programme starts with 7B, then moves to each larger
-size class and to other model families. At every class the model must improve on every axis that matters (reasoning,
-mathematics, code, instruction following and knowledge) without a material regression against its base. Each class
-is complete only when no public fine-tune improves any axis further.
+---
 
-## The process
+## Open Questions
 
-The process is defined once and run the same way at every class, so every release can be rebuilt and re-measured.
+- Does consolidation hold on every axis at once, across a full board?
+- Does it beat ensembling, routing and best-of-n sampling at matched compute?
+- Does it carry to larger classes and other model families?
+- Does it hold on calibration, hallucination and instruction following?
 
-1. **Start** from a strong open base model in the class.
-2. **Discover** the compatible fine-tunes the community has published for it.
-3. **Assimilate** them into one model.
-4. **Measure** every axis against the base model on identical items, in the same session, on public benchmark protocols.
-5. **Release** a model only when it holds against its base and the previous release on the whole board, and publish every regression next to every gain, with per-item evidence. This is the release rule from the corrected releases on; v1 and v2 predate it, and their cards list their regressions.
-6. **Repeat** until nothing more improves, then carry the process to the next class.
+Each answer is published as it is measured.
 
-The construction method is proprietary. The evaluation is fully public: protocols, per-item outputs and the tools
-that recompute every number.
+---
 
-## Where the programme stands
+## Who It Is For
 
-| Release | Status |
-|---|---|
-| [converge v1](https://huggingface.co/Optitransfer/Qwen2.5-7B-Instruct-converge-collective-v1) | Public. The first 7B release, on Qwen2.5-7B-Instruct. It carries a merge defect that has since been corrected; see the notice on its card. |
-| [converge v2](https://huggingface.co/Optitransfer/Qwen2.5-7B-Instruct-converge-collective-v2) | Public. The current release, with a full validation board. It is built on v1, so it inherits the defect. |
-| corrected v1 and v2 | Built with the same recipes and the defect removed. Being benchmarked across the full board now. |
-| converge v3 | Next. Released only once it holds against its base model and the previous release on the whole board. |
+| | |
+|:--|:--|
+| **Open community** | A stronger open model per class, with the evidence to check it |
+| **Regulated industries** | Healthcare and finance teams whose fine-tunes cannot leave their boundary, and who must show how each model change was made |
+| **National programmes** | Capability that grows by contribution, without dependence on a few vendors |
 
-## How we report
+**Open model, paid guarantees.** The public model stays open. The planned commercial offering is private consolidation of an organisation's own fine-tunes, inside its boundary, with the same measurement and evidence.
 
-- **Paired, never selected.** Every score is the model minus its base on the same items, with a 95% confidence
-  interval and an exact significance test.
-- **Regressions are first-class.** Each card lists what went down as prominently as what went up.
-- **Reproducible only.** Public numbers come from reproducible runs on public protocols: HELM-protocol GSM8K and
-  MMLU, MMLU-Pro (TIGER-Lab), ZeroEval, EvalPlus HumanEval+ and MBPP+, MATH, AIME, ARC and IFEval.
-- **Corrected in public.** When a defect is found, the affected cards say so first.
+---
 
-## Who it is for
+## Demonstrations
 
-- **The open community.** A stronger open model at each class, with the evidence to check it.
-- **Healthcare.** A hospital network fine-tunes models on patient data that cannot leave its walls. Converge brings
-  those internal fine-tunes together inside its boundary, with a record of every change.
-- **Finance.** A regulated bank must show auditors how each model update was made, what changed and that nothing
-  regressed. Converge's per-item record is that audit trail.
-- **Sovereign AI.** A national programme needs frontier capability without depending on a few foreign vendors.
-  Converge grows by contribution, not by procurement.
+| Space | Description |
+|:--|:--|
+| [crdt-merge](https://huggingface.co/spaces/Optitransfer/crdt-merge) | Merge real models with full provenance |
+| [Convergence Lab](https://huggingface.co/spaces/Optitransfer/convergence-lab) | 26 strategies, convergence visualisation, experiments |
+| [Data Playground](https://huggingface.co/spaces/Optitransfer/crdt-merge-data) | CRDT merging on tabular data |
+| [Federation](https://huggingface.co/spaces/Optitransfer/crdt-merge-federation) | Multi-node gossip convergence |
 
-## How it is offered
+---
 
-**A free model, with paid guarantees.** The public model stays open, and the free tier is never limited to sell a
-paid one. Around it, the first commercial product is auditable private aggregation: Optitransfer brings an
-organisation's own fine-tunes together inside its boundary, with the same measurement and evidence. Compliance
-evidence, sovereign deployments and private federations follow.
+## Open Components and Papers
 
-## Open components
+| | |
+|:--|:--|
+| [crdt-merge](https://github.com/mgillr/crdt-merge) | Conflict-free (CRDT) merging for data, models and agents, with the E4 trust-scoring engine. `pip install crdt-merge` |
+| [acfa-rs](https://github.com/mgillr/acfa-rs) | Accountable federated aggregation with signed, offline-verifiable receipts (Apache-2.0) |
+| [arXiv:2605.19373](https://arxiv.org/abs/2605.19373) | Conflict-Free Replicated Data Types for Neural Network Model Merging |
+| [arXiv:2607.10305](https://arxiv.org/abs/2607.10305) | Byzantine Accountability Without Consensus |
 
-- **[crdt-merge](https://github.com/mgillr/crdt-merge):** conflict-free (CRDT) merging for data, JSON, models and
-  agents, with the E4 trust-scoring engine ([paper](https://arxiv.org/abs/2605.19373)).
-- **[acfa-rs](https://github.com/mgillr/acfa-rs):** accountable federated aggregation with signed, offline-verifiable
-  receipts (Apache-2.0, [arXiv:2607.10305](https://arxiv.org/abs/2607.10305)).
+---
 
 ## Team
 
-**Ryan Gillespie**, founder, architect and operator · [@Optitransfer](https://huggingface.co/Optitransfer)
+**Ryan Gillespie**, founder · [@Optitransfer](https://huggingface.co/Optitransfer)
+
+Fifteen years of verification and validation on safety-critical systems: railway signalling (Crossrail, ETCS), medical devices taken to CE mark (Roche) and aviation (Lilium). Converge applies that discipline to model weights.
 
-- Fifteen years of verification and validation on safety-critical systems: railway signalling (Crossrail, ETCS),
-  medical devices taken to CE mark (Roche), and aviation (Lilium). Converge's evidence-first design is that
-  discipline applied to model weights.
-- Built the full stack, from the merge protocol and its CRDT substrate to the evaluation and evidence system.
-- Ran the research programme behind every published measurement.
-- Author of the paper on CRDT-based model merging ([arXiv:2605.19373](https://arxiv.org/abs/2605.19373)).
+---
+
+<div align="center">
 
-<sub>Current releases are licensed for non-commercial research, a condition inherited from an upstream source model.
-Each model card states its licence in full.</sub>
+Switzerland · [data@optitransfer.ch](mailto:data@optitransfer.ch)
+
+<sub>Current model releases are licensed for non-commercial research, a condition inherited from an upstream source model. Each model card states its licence in full.</sub>
+
+</div>

index.html

@@ -1,67 +1,350 @@
-<!doctype html>
+<!DOCTYPE html>
 <html lang="en">
 <head>
-<meta charset="utf-8">
-<meta name="viewport" content="width=device-width, initial-scale=1">
-<title>Optitransfer · converge</title>
-<meta name="description" content="Converge: building frontier-level open models from the work the open-source community has already published. One repeatable process at every model class, with every result published with per-item evidence.">
+<meta charset="utf-8" />
+<meta name="viewport" content="width=device-width, initial-scale=1.0" />
+<meta name="description" content="Optitransfer: open models built to place at the top of their class on public leaderboards, from the work the open-source community has already published. One open model per class, every result measured against its base and published with its evidence." />
+<meta name="keywords" content="open-weights LLM, model merging, capability consolidation, Qwen2.5, 7B, reproducible evaluation, benchmarks, leaderboards, crdt-merge" />
+<meta name="author" content="Ryan Gillespie" />
+<meta property="og:title" content="Optitransfer" />
+<meta property="og:description" content="Open models built to place at the top of their class on public leaderboards, from the work the open-source community has already published." />
+<meta property="og:type" content="website" />
+<meta property="og:url" content="https://huggingface.co/OptiTransferData" />
+<title>Optitransfer</title>
 <style>
-  :root { --bg: #ffffff; --fg: #1a1d24; --muted: #5b6270; --line: #e6e8ec; --accent: #3b4fd8; }
-  @media (prefers-color-scheme: dark) { :root { --bg: #0f1115; --fg: #e8eaf0; --muted: #9aa1ae; --line: #252a33; --accent: #8ea0ff; } }
-  * { box-sizing: border-box; }
-  body { margin: 0; background: var(--bg); color: var(--fg); font: 16px/1.6 system-ui, -apple-system, "Segoe UI", sans-serif; }
-  main { max-width: 760px; margin: 0 auto; padding: 56px 16px 72px; }
-  h1 { font-size: 2.2rem; margin: 0 0 .25rem; letter-spacing: -.01em; }
-  .lede { font-size: 1.15rem; margin: 0 0 .25rem; }
-  .sub { color: var(--muted); margin: 0 0 2rem; }
-  h2 { font-size: 1.05rem; text-transform: uppercase; letter-spacing: .06em; color: var(--muted); margin: 2.25rem 0 .5rem; }
-  a { color: var(--accent); text-decoration: none; } a:hover { text-decoration: underline; }
-  ul, ol { padding-left: 1.2rem; } li { margin: .3rem 0; }
-  .rel { border-top: 1px solid var(--line); }
-  .rel div { display: flex; gap: 1rem; padding: .6rem 0; border-bottom: 1px solid var(--line); }
-  .rel b { min-width: 9.5rem; flex-shrink: 0; }
-  footer { margin-top: 3rem; color: var(--muted); font-size: .9rem; }
-  @media (max-width: 520px) { .rel div { flex-direction: column; gap: .1rem; } }
+  :root {
+    --bg: #0a0a0a;
+    --surface: #111113;
+    --border: #1e1e22;
+    --text: #e4e4e7;
+    --muted: #71717a;
+    --accent: #a1a1aa;
+    --mono: 'SF Mono', 'Cascadia Code', 'Fira Code', Consolas, monospace;
+    --sans: -apple-system, BlinkMacSystemFont, 'Segoe UI', system-ui, sans-serif;
+  }
+  * { margin: 0; padding: 0; box-sizing: border-box; }
+  body {
+    font-family: var(--sans);
+    background: var(--bg);
+    color: var(--text);
+    line-height: 1.6;
+    -webkit-font-smoothing: antialiased;
+  }
+  .container {
+    max-width: 720px;
+    margin: 0 auto;
+    padding: 4rem 2rem;
+  }
+  .header {
+    text-align: center;
+    margin-bottom: 4rem;
+  }
+  .header h1 {
+    font-size: 1.75rem;
+    font-weight: 500;
+    letter-spacing: 0.08em;
+    text-transform: uppercase;
+    color: var(--accent);
+    margin-bottom: 1rem;
+  }
+  .header .tagline {
+    font-size: 1.05rem;
+    color: var(--muted);
+    font-weight: 400;
+    max-width: 480px;
+    margin: 0 auto;
+  }
+  .header .install {
+    margin-top: 2rem;
+    display: inline-block;
+    font-family: var(--mono);
+    font-size: 0.875rem;
+    color: var(--text);
+    background: var(--surface);
+    border: 1px solid var(--border);
+    padding: 0.6rem 1.5rem;
+    border-radius: 6px;
+  }
+  .badges {
+    display: flex;
+    justify-content: center;
+    gap: 0.5rem;
+    flex-wrap: wrap;
+    margin-top: 1.5rem;
+  }
+  .badges a { opacity: 0.85; transition: opacity 0.2s; }
+  .badges a:hover { opacity: 1; }
+  .section { margin-bottom: 3rem; }
+  .section h2 {
+    font-size: 0.75rem;
+    font-weight: 600;
+    letter-spacing: 0.12em;
+    text-transform: uppercase;
+    color: var(--muted);
+    margin-bottom: 1.25rem;
+    padding-bottom: 0.5rem;
+    border-bottom: 1px solid var(--border);
+  }
+  .stats {
+    display: grid;
+    grid-template-columns: repeat(4, 1fr);
+    gap: 1rem;
+    text-align: center;
+  }
+  .stat .number {
+    font-family: var(--mono);
+    font-size: 1.5rem;
+    font-weight: 600;
+    color: var(--text);
+  }
+  .stat .label {
+    font-size: 0.7rem;
+    font-weight: 500;
+    letter-spacing: 0.08em;
+    text-transform: uppercase;
+    color: var(--muted);
+    margin-top: 0.25rem;
+  }
+  .links {
+    display: grid;
+    grid-template-columns: repeat(auto-fit, minmax(200px, 1fr));
+    gap: 1rem;
+  }
+  .link-card {
+    background: var(--surface);
+    border: 1px solid var(--border);
+    border-radius: 8px;
+    padding: 1.25rem;
+    text-decoration: none;
+    color: var(--text);
+    transition: border-color 0.2s, transform 0.15s;
+    display: block;
+  }
+  .link-card:hover {
+    border-color: var(--accent);
+    transform: translateY(-1px);
+  }
+  .link-card .label {
+    font-size: 0.7rem;
+    font-weight: 600;
+    letter-spacing: 0.1em;
+    text-transform: uppercase;
+    color: var(--muted);
+    margin-bottom: 0.35rem;
+  }
+  .link-card .value { font-size: 0.95rem; }
+  .spaces { display: grid; gap: 0.75rem; }
+  .space {
+    display: flex;
+    justify-content: space-between;
+    align-items: center;
+    background: var(--surface);
+    border: 1px solid var(--border);
+    border-radius: 8px;
+    padding: 1rem 1.25rem;
+    text-decoration: none;
+    color: var(--text);
+    transition: border-color 0.2s;
+  }
+  .space:hover { border-color: var(--accent); }
+  .space .name { font-weight: 500; font-size: 0.95rem; }
+  .space .desc { font-size: 0.8rem; color: var(--muted); }
+  .space .arrow { color: var(--muted); font-size: 1.1rem; }
+  .research-list { list-style: none; padding: 0; }
+  .research-list li {
+    color: var(--muted);
+    font-size: 0.9rem;
+    padding: 0.4rem 0;
+    border-bottom: 1px solid var(--border);
+  }
+  .research-list li:last-child { border-bottom: none; }
+  .research-list li::before {
+    content: '';
+    display: inline-block;
+    width: 6px;
+    height: 6px;
+    border-radius: 50%;
+    background: var(--accent);
+    margin-right: 0.75rem;
+    vertical-align: middle;
+  }
+  .footer {
+    text-align: center;
+    color: var(--muted);
+    font-size: 0.8rem;
+    margin-top: 2rem;
+    padding-top: 2rem;
+    border-top: 1px solid var(--border);
+  }
+  .footer a { color: var(--accent); text-decoration: none; }
+  .footer a:hover { text-decoration: underline; }
+  @media (max-width: 600px) {
+    .container { padding: 2rem 1rem; }
+    .stats { grid-template-columns: repeat(2, 1fr); }
+    .links { grid-template-columns: 1fr; }
+  }
+  .section p { color: var(--muted); font-size: 0.95rem; margin-bottom: 0.75rem; }
+  .section p strong { color: var(--text); font-weight: 500; }
+  .stats.three { grid-template-columns: repeat(3, 1fr); }
+  .research-list li strong { color: var(--text); font-weight: 500; }
 </style>
 </head>
 <body>
-<main>
-  <h1>Optitransfer</h1>
-  <p class="lede"><strong>Converge:</strong> building frontier-level open models from the work the open-source community has already published.</p>
-  <p class="sub">One repeatable process at every model class. Every result is measured against its base model and published with per-item evidence.</p>
+<div class="container">
 
-  <h2>Purpose</h2>
-  <p>The community has published thousands of fine-tuned models, each adding some skill to a shared base. Converge brings that work together into a single open model for each class, without pre-training a new one. It measures every change, keeps every gain and publishes every regression.</p>
+  <header class="header">
+    <h1>Optitransfer</h1>
+    <p class="tagline">
+      Open models built to place at the top of their class on public leaderboards, from the work the open-source community has already published.
+    </p>
+    <div class="badges">
+      <a href="https://huggingface.co/Optitransfer"><img src="https://img.shields.io/badge/models-converge%20v1%20%C2%B7%20v2-blue" alt="Models" /></a>
+      <a href="https://arxiv.org/abs/2605.19373"><img src="https://img.shields.io/badge/arXiv-2605.19373-b31b1b" alt="Paper" /></a>
+      <a href="https://arxiv.org/abs/2607.10305"><img src="https://img.shields.io/badge/arXiv-2607.10305-b31b1b" alt="Paper" /></a>
+      <a href="https://pypi.org/project/crdt-merge/"><img src="https://img.shields.io/pypi/v/crdt-merge?label=crdt-merge&color=blue" alt="crdt-merge" /></a>
+    </div>
+  </header>
 
-  <h2>The goal</h2>
-  <p>Reach frontier-level performance in every model class: 7B first, then each larger size class and other model families. The model must improve on every axis without a material regression against its base, until no public fine-tune improves any axis further.</p>
+  <div class="section">
+    <h2>At a Glance</h2>
+    <div class="stats">
+      <div class="stat"><div class="number">7B</div><div class="label">Class in progress</div></div>
+      <div class="stat"><div class="number">2</div><div class="label">Public releases</div></div>
+      <div class="stat"><div class="number">12</div><div class="label">Board benchmarks</div></div>
+      <div class="stat"><div class="number">2</div><div class="label">Papers</div></div>
+    </div>
+  </div>
+
+  <div class="section">
+    <h2>The Thesis</h2>
+    <p>Thousands of fine-tuned models sit on public repositories. Each adds a skill to a shared base model. Their training is already paid for.</p>
+    <p><strong>Hypothesis:</strong> that capability can be consolidated into one model per class, without pre-training and without trading one skill for another. Converge is the programme that tests it, one release at a time, in public.</p>
+  </div>
+
+  <div class="section">
+    <h2>Where We Are</h2>
+    <ul class="research-list">
+      <li><strong>v2 against its base.</strong> 12 benchmarks: 2 improved, 2 regressed, 8 with no measurable difference.</li>
+      <li><strong>Defect found.</strong> In the merge that produced v1, inherited by v2. Corrected, and being re-benchmarked.</li>
+      <li><strong>Not yet shown.</strong> A release that improves on its base with no regression. The next release must pass that test.</li>
+    </ul>
+  </div>
+
+  <div class="section">
+    <h2>Why It Matters</h2>
+    <ul class="research-list">
+      <li><strong>Science.</strong> Whether consolidation can hold on every axis at once is an open, testable question.</li>
+      <li><strong>Economics.</strong> If the hypothesis holds, the marginal cost of a stronger open model is evaluation compute, not pre-training.</li>
+      <li><strong>Adoption.</strong> Capability without dependence on one vendor, with a record of every model change.</li>
+    </ul>
+  </div>
 
-  <h2>The process</h2>
-  <ol>
-    <li><strong>Start</strong> from a strong open base model in the class.</li>
-    <li><strong>Discover</strong> the compatible fine-tunes the community has published.</li>
-    <li><strong>Assimilate</strong> them into one model.</li>
-    <li><strong>Measure</strong> every axis against the base on identical items, on public protocols.</li>
-    <li><strong>Release</strong> a model only when it holds on the whole board, and publish every regression. This is the release rule from the corrected releases on.</li>
-    <li><strong>Repeat</strong>, then carry the process to the next class.</li>
-  </ol>
-  <p>The construction method is proprietary. The evaluation is fully public.</p>
+  <div class="section">
+    <h2>The Goal</h2>
+    <p><strong>Top placements in class, across the public leaderboards.</strong> 7B first, then each larger size class and other model families. Every axis must improve with no statistically significant regression against the base, until no public fine-tune improves any axis further.</p>
+  </div>
+
+  <div class="section">
+    <h2>The Process</h2>
+    <ul class="research-list">
+      <li><strong>Start</strong> from a strong open base model in the class</li>
+      <li><strong>Discover</strong> the compatible fine-tunes the community has published</li>
+      <li><strong>Assimilate</strong> them into one model</li>
+      <li><strong>Measure</strong> every axis against the base, on identical items and public protocols</li>
+      <li><strong>Release</strong> only when the model holds on the whole board</li>
+      <li><strong>Repeat</strong>, then carry the process to the next class</li>
+    </ul>
+  </div>
+
+  <div class="section">
+    <h2>Research Releases</h2>
+    <p>Releases are published under this organisation from converge v3 onwards. The research that led to them is on the founder's page.</p>
+    <div class="spaces">
+      <a href="https://huggingface.co/Optitransfer/Qwen2.5-7B-Instruct-converge-collective-v2" class="space">
+        <div>
+          <div class="name">converge v2</div>
+          <div class="desc">The current release, with a 12-benchmark validation board</div>
+        </div>
+        <span class="arrow">&rarr;</span>
+      </a>
+      <a href="https://huggingface.co/Optitransfer/Qwen2.5-7B-Instruct-converge-collective-v1" class="space">
+        <div>
+          <div class="name">converge v1</div>
+          <div class="desc">The first 7B release; see the correction notice on its card</div>
+        </div>
+        <span class="arrow">&rarr;</span>
+      </a>
+    </div>
+  </div>
+
+  <div class="section">
+    <h2>Demonstrations</h2>
+    <div class="spaces">
+      <a href="https://huggingface.co/spaces/Optitransfer/crdt-merge" class="space">
+        <div>
+          <div class="name">crdt-merge</div>
+          <div class="desc">Merge real models with full provenance</div>
+        </div>
+        <span class="arrow">&rarr;</span>
+      </a>
+      <a href="https://huggingface.co/spaces/Optitransfer/convergence-lab" class="space">
+        <div>
+          <div class="name">Convergence Lab</div>
+          <div class="desc">26 strategies, convergence visualisation, experiments</div>
+        </div>
+        <span class="arrow">&rarr;</span>
+      </a>
+      <a href="https://huggingface.co/spaces/Optitransfer/crdt-merge-data" class="space">
+        <div>
+          <div class="name">Data Playground</div>
+          <div class="desc">CRDT merging on tabular data</div>
+        </div>
+        <span class="arrow">&rarr;</span>
+      </a>
+      <a href="https://huggingface.co/spaces/Optitransfer/crdt-merge-federation" class="space">
+        <div>
+          <div class="name">Federation</div>
+          <div class="desc">Multi-node gossip convergence</div>
+        </div>
+        <span class="arrow">&rarr;</span>
+      </a>
+    </div>
+  </div>
 
-  <h2>Releases</h2>
-  <div class="rel">
-    <div><b><a href="https://huggingface.co/Optitransfer/Qwen2.5-7B-Instruct-converge-collective-v1">converge v1</a></b><span>Public. The first 7B release. It carries a corrected merge defect; see its card.</span></div>
-    <div><b><a href="https://huggingface.co/Optitransfer/Qwen2.5-7B-Instruct-converge-collective-v2">converge v2</a></b><span>Public. The current release. It is built on v1, so it inherits the defect.</span></div>
-    <div><b>corrected v1 and v2</b><span>Same recipes, defect removed. Being benchmarked across the full board now.</span></div>
-    <div><b>converge v3</b><span>Next. Released only once it holds against its base and the previous release.</span></div>
+  <div class="section">
+    <h2>Links</h2>
+    <div class="links">
+      <a href="https://huggingface.co/Optitransfer" class="link-card">
+        <div class="label">Models</div>
+        <div class="value">huggingface.co/Optitransfer</div>
+      </a>
+      <a href="https://github.com/mgillr/crdt-merge" class="link-card">
+        <div class="label">Source</div>
+        <div class="value">github.com/mgillr/crdt-merge</div>
+      </a>
+      <a href="https://github.com/mgillr/acfa-rs" class="link-card">
+        <div class="label">Source</div>
+        <div class="value">github.com/mgillr/acfa-rs</div>
+      </a>
+      <a href="https://arxiv.org/abs/2605.19373" class="link-card">
+        <div class="label">Paper</div>
+        <div class="value">arXiv:2605.19373</div>
+      </a>
+      <a href="https://arxiv.org/abs/2607.10305" class="link-card">
+        <div class="label">Paper</div>
+        <div class="value">arXiv:2607.10305</div>
+      </a>
+      <a href="https://pypi.org/project/crdt-merge/" class="link-card">
+        <div class="label">Package</div>
+        <div class="value">pip install crdt-merge</div>
+      </a>
+    </div>
   </div>
 
-  <h2>Open components</h2>
-  <ul>
-    <li><a href="https://github.com/mgillr/crdt-merge">crdt-merge</a>: conflict-free merging for data, models and agents, with the E4 trust engine.</li>
-    <li><a href="https://github.com/mgillr/acfa-rs">acfa-rs</a>: accountable federated aggregation with offline-verifiable receipts.</li>
-  </ul>
+  <footer class="footer">
+    Ryan Gillespie, founder &middot; Switzerland &middot; <a href="mailto:data@optitransfer.ch">data@optitransfer.ch</a>
+  </footer>
 
-  <footer>Ryan Gillespie, founder · <a href="https://huggingface.co/Optitransfer">@Optitransfer</a></footer>
-</main>
+</div>
 </body>
 </html>