|
Download README.md from MouseWurX/REAP-scores-28Bob-Riker: direct link, hf CLI and curl.
- Browser
- Download file 6.99 kB
-
https://huggingface.co/datasets/MouseWurX/REAP-scores-28Bob-Riker/resolve/main/README.md
- Command line
-
hf download hf://datasets/MouseWurX/REAP-scores-28Bob-Riker/README.md
-
curl -L -o README.md https://huggingface.co/datasets/MouseWurX/REAP-scores-28Bob-Riker/resolve/main/README.md
6.99 kB
| license: cc-by-sa-4.0 | |
| tags: | |
| - reap | |
| - moe | |
| - expert-pruning | |
| - mixture-of-experts | |
| - bob | |
| # REAP scores: 28Bob "Riker" | |
| Riker is one of Bob's own brains: Qwen3.6-35B-A3B cut to 208 experts per layer and healed on Bob's lessons. This page holds its experts, scored and ranked by REAP, and the keep lists that were tested, in **one file**: | |
| `28Bob-Riker.json`. Use it to run the model with fewer experts **without cutting the model file**. Made for | |
| [Bob](https://github.com/MouseWurX/Bob), the open-source desktop assistant. | |
| Bob and his brains' names are inspired by Dennis E. Taylor's novels (not affiliated). | |
| ## What REAP is | |
| REAP (Router-weighted Expert Activation Pruning, Cerebras 2025) scores each expert by **router weight × the size of the | |
| expert's output**, summed over every word it was picked for. Here the sums run over **100 of Bob's training lessons | |
| (270,267 tokens)**, with the model served at 12 experts per word from MouseWurX/28Bob-Riker: 28Bob-Q4_K_M.gguf. Experts that are picked rarely but | |
| matter a lot when they are keep a high score; counting picks alone would miss them. | |
| ## The file | |
| `28Bob-Riker.json`: `model`, `base`, `method`, `lessons`, `tokens`, `layers` (40), `experts` (208 per layer), `experts_per_word` (12), | |
| `scores.picks` and `scores.mass` (one row per layer, one number per expert), `experts_by_layer` (each layer's expert ids, | |
| strongest first), `keep_lists` (by percent kept: one list of expert ids per layer; `null` = every expert on), `results` (the | |
| held-out score at each step) and `picked`. | |
| ## Using a keep list (no file cut) | |
| Brain Engine (Bob's llama.cpp build with the expert-mask patch) switches experts on and off live: | |
| ```bash | |
| # keep only the listed experts in each layer (here the 35% list) | |
| jq '{keep: .keep_lists["35"]}' 28Bob-Riker.json | curl -X POST http://127.0.0.1:8080/expert-mask -H "Content-Type: application/json" -d @- | |
| # every expert back on | |
| curl -X POST http://127.0.0.1:8080/expert-mask -H "Content-Type: application/json" -d '{"keep": null}' | |
| ``` | |
| A mask changes which experts the router may pick; it doesn't free memory by itself. To save memory, load only the kept | |
| experts (for example, keep the rest in system RAM), or cut a file from the same list. | |
| ## Held-out results | |
| Judged on Bob's 196 held-out lessons (314 assistant turns, never trained on), thinking off. "Right" means the same tools and | |
| arguments as the full Qwen3.6-35B-A3B teacher, or correctly no tool. | |
| | Kept per layer | Share | Right tool | Right arguments | Correctly no tool | Held-out overall | Tokens/s | | |
| |---|---|---|---|---|---|---| | |
| | 208 of 208 | 100% | 70/123 | 47/123 | 163/191 | **210/314** (66.9%) | 114.4 | | |
| | 156 of 208 | 75% | 66/123 | 45/123 | 159/191 | **204/314** (65.0%) | 94.2 | | |
| | 104 of 208 | 50% | 64/123 | 32/123 | 131/191 | **163/314** (51.9%) | 97.7 | | |
| | 73 of 208 | 35% | 66/123 | 33/123 | 39/191 | **72/314** (22.9%) | 92.5 | | |
| The calibration's own pick (smallest step within 2 points of the full score): **75%**. Read it with the table: 75% is 6 points down. | |
| ## Part of Bob | |
| This is part of **[Bob](https://github.com/MouseWurX/Bob)**, a free, open-source AI agent that runs on your own computer (Windows, Mac, Linux, Raspberry Pi, Android). Nothing goes online unless you turn it on. | |
| **What Bob does** | |
| - **A real agent:** researches with sources, writes, runs and fixes code, reads and edits documents and files, works other apps, and makes pictures, clips, music and a voice. | |
| - **His own brains:** models cut and retrained to run Bob on everyday graphics cards, with every score, lesson set and dollar published. | |
| - **Bob-Net:** your Bobs share graphics cards and memory and check each other's work, on your home network. | |
| - **Gaming friendly:** the brain unloads when idle, and a Gaming session gives the graphics card back to your game. | |
| - **Private:** no account, no telemetry. Everything optional starts switched off. | |
| **The Computer Science Lab** ([guide](https://github.com/MouseWurX/Bob/blob/main/app/docs/USER-GUIDE.md#make-your-own-brain-the-lab), [code](https://github.com/MouseWurX/Bob/tree/main/app/model-lab)), where this was made and measured. It's a workbench inside Bob, with no command line: | |
| - **Make a brain:** shrink it, cut experts, or take out sight, hearing or the refusal habit. A brain is usable only after it has been tested. | |
| - **Forge:** every way to make a new brain, side by side. It writes the plan for jobs that need a rented card. | |
| - **Test:** scored challenges, one brain in six sizes, long-context and speed tests, and blind taste tests. | |
| - **Teach:** teach a brain a lesson set, or have a bigger brain teach a smaller one. | |
| - **Look inside:** see which experts your work uses. **fNRI** scans a brain while it answers and plays it back in 3D. | |
| **Training protocols** ([Forge guide](https://github.com/MouseWurX/Bob/blob/main/app/model-lab/forge/README.md), [Forge journal](https://github.com/MouseWurX/Bob/blob/main/app/docs/FORGE-JOURNAL.md), [Making smaller models](https://github.com/MouseWurX/Bob/blob/main/app/docs/MAKING-SMALLER-MODELS.md)). Every model goes through the same steps, in order: baseline → clean first → analyse → train → cut in steps → heal → check every target → learn from its own answers → blend → live dials → publish. Each step is benchmarked. Why we chose this order: | |
| - **Baseline first,** so every change is measured against the untouched model, not against a guess. | |
| - **Clean before teaching, cut after training,** so a cut keeps what training taught and no fault gets baked in. | |
| - **Measure before cutting (REAP),** so the experts that do the least work are the ones removed, per job, from real traces. | |
| - **Cut in steps and heal from the full model's odds** (distillation), because small steps show where quality breaks, and copying the teacher's odds recovers more than plain answers can. | |
| - **No target may drop,** so a gain in tools can't quietly cost coding, facts or safety. | |
| - **Learn from its own answers** (on-policy, reward for tool calls that really work), because a model fixes the mistakes it actually makes, not the ones a teacher imagines. | |
| - **Honest data:** test items are never taught, surgery and evil-check data never train, and everything is scrubbed of private data first. | |
| **The plan:** what comes next, and why, in the [roadmap](https://github.com/MouseWurX/Bob/pull/230). | |
| **Costs:** every rental, its card, its hours and what it bought, on the [costs page](https://github.com/MouseWurX/Bob/blob/main/app/docs/COSTS.md). | |
| ## Credit | |
| Base model: Qwen3.6-35B-A3B by Alibaba's Qwen team (Apache 2.0, https://huggingface.co/Qwen/Qwen3.6-35B-A3B). The scores and keep lists here are CC BY-SA 4.0. | |
| ## Other REAP scores | |
| - [35Bob "Bobert"](https://huggingface.co/datasets/MouseWurX/REAP-scores-35Bob-Bobert) | |
| - [gpt-oss-120b](https://huggingface.co/datasets/MouseWurX/REAP-scores-gpt-oss-120b) | |
| - Bob's lessons, and the answers from this benchmark run: [MouseWurX/bob-lessons](https://huggingface.co/datasets/MouseWurX/bob-lessons) | |