Upload README.md with huggingface_hub

README.md

@@ -1,61 +1,40 @@
 ---
 license: other
-language:
-- en
-tags:
-- midtraining
-- code
-- agentic
-- software-engineering
-- multi-agent
-task_categories:
-- text-generation
-configs:
-- config_name: default
-  data_files:
-  - split: train
-    path: data/train-*
-dataset_info:
-  features:
-  - name: text
-    dtype: string
-  - name: source
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 1610042422
-    num_examples: 727337
-  download_size: 806246956
-  dataset_size: 1610042422
+language: [en]
+tags: [midtraining, code, agentic, software-engineering, multi-agent]
+task_categories: [text-generation]
 ---
 
 # CooperData v3 — Midtraining Blend (Qwen3.5-9B cooperative SWE agents)
 
-All-token **midtraining** mixture that bridges `Qwen/Qwen3.5-9B` (instruct) toward the
-cooperative multi-agent SWE-coding SFT distribution. **One document per row** (`text`, tagged by
-`source`) — NOT packed — so `trl.SFTTrainer(packing=False)` tokenizes per-doc and the Gated-DeltaNet
-recurrence stays per-document. **~390M tokens.**
+All-token **midtraining** mixture that bridges `Qwen/Qwen3.5-9B` (instruct) toward the cooperative
+multi-agent SWE-coding SFT distribution. **One document per row** (`text`, tagged by `source`) — NOT
+packed — so `trl.SFTTrainer(packing=False)` tokenizes per-doc and the Gated-DeltaNet recurrence
+stays per-document. **~390M tokens.**
 
 ## Composition
-| source | tokens | share |
-|---|---|---|
-| `web` | 210.0M | 54% |
-| `math` | 55.0M | 14% |
-| `instruct` | 25.0M | 6% |
-| `coop` | 80.0M | 21% |
-| `agentic` | 20.0M | 5% |
+| source | tokens | share | role |
+|---|---|---|---|
+| `web` | 210.0M | 54% | general |
+| `math` | 55.0M | 14% | general |
+| `instruct` | 25.0M | 6% | general |
+| `coop` | 60.0M | 15% | bridge |
+| `swe_smith` | 15.0M | 4% | bridge |
+| `nebius` | 15.0M | 4% | bridge |
+| `social` | 10.0M | 3% | bridge |
 
-- **General (~81%)** — anti-forgetting bulk + the distant domains (code/math) midtraining helps most:
-  `web`/`math`/`instruct` = [allenai/dolmino-mix-1124](https://huggingface.co/datasets/allenai/dolmino-mix-1124) (`dclm`/`math`/`flan`), OLMo-2's curated midtraining mix.
-- **Bridge (~19%, coop-dominant)** — agentic data resembling the SFT target:
-  - `coop`: our cooperative multi-agent SWE trajectories (mini-swe-agent, **structured tool-calling**), rendered with the model's **real** chat template → native `<tool_call>`/`<tool_response>` (matches the `qwen3_coder` serving parser). This also carries the "good collaborator" signal — the agents coordinate as colleagues on a shared codebase.
-  - `agentic`: [ricdomolm/mini-coder-trajs-400k](https://huggingface.co/datasets/ricdomolm/mini-coder-trajs-400k), verified rollouts (bash-in-content style — midtrain *exposure* only; the SFT stage uses coop tool-calling exclusively).
+General **~74%** / bridge **~26%** (coop-dominant). Agentic/social sources are rendered with
+the model's **real** chat template (so coop tool-calls become native `<tool_call>`/`<tool_response>`,
+matching the `qwen3_coder` serving parser).
 
 ## Methodology
-Grounded in ["Midtraining Bridges Pretraining and Posttraining Distributions" (2510.14865)](https://arxiv.org/abs/2510.14865): because specialized data is introduced *late* (a finished instruct model), the conservative-weight finding (high specialized weight late → catastrophic forgetting) dictates a **general-dominant** mix (~81% general / ~19% bridge). Heavy agentic specialization is left to SFT.
+Grounded in ["Midtraining Bridges Pretraining and Posttraining Distributions" (2510.14865)](https://arxiv.org/abs/2510.14865):
+specialized data is introduced *late* (a finished instruct model), so the conservative-weight finding
+(high specialized weight late -> catastrophic forgetting) dictates a **general-dominant** mix. The
+bridge is coop-dominant — near-domain agentic-coding on a coding model (low forgetting risk) and the
+exact SFT target. Heavy agentic specialization is left to SFT.
 
-## Loss / usage
-All-token (full LM loss), `packing=False`:
+## Usage (all-token, packing=False)
 ```python
 from trl import SFTTrainer, SFTConfig
 from datasets import load_dataset
@@ -63,4 +42,7 @@ ds = load_dataset("CooperBench/cooperdata-v3-midtrain-blend", split="train")
 SFTTrainer(model="Qwen/Qwen3.5-9B", train_dataset=ds,
            args=SFTConfig(dataset_text_field="text", packing=False, max_length=8192))
 ```
-Inherits the licenses of its constituent datasets.
+
+## Build
+`coopertrain/train/modal/datamix/` (sources + mixture + cleaning). Inherits the licenses of its
+constituent datasets (Dolmino, mini-coder-trajs, OdysSim) + our coop trajectories.