Upload folder using huggingface_hub

.gitattributes

@@ -57,3 +57,10 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 # Video files - compressed
 *.mp4 filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
+apienv/dedeucerl_apienv_n7_e7_b200_train.json filter=lfs diff=lfs merge=lfs -text
+apienv/hf_apienv_n7_e7_b200_train.jsonl filter=lfs diff=lfs merge=lfs -text
+exprpolicy/dedeucerl_exprpolicy_b60_train.json filter=lfs diff=lfs merge=lfs -text
+exprpolicy/hf_exprpolicy_b60_train.jsonl filter=lfs diff=lfs merge=lfs -text
+mealy/dedeucerl_mealy_n4_b100_train.json filter=lfs diff=lfs merge=lfs -text
+protocol/dedeucerl_protocol_n4_e5_b120_train.json filter=lfs diff=lfs merge=lfs -text
+protocol/hf_protocol_n4_e5_b120_train.jsonl filter=lfs diff=lfs merge=lfs -text

GENERATION.json

@@ -1,24 +1,109 @@
 {
   "format": "DedeuceRL HF export",
   "note": "HF_DATA is local-only and gitignored. Upload hf_*.jsonl to Hugging Face dataset repo.",
+  "generated_at": "2025-01-29",
   "splits": [
     {
       "skin": "mealy",
       "subset": "test",
+      "seeds": "0-99",
+      "params": {
+        "n_states": 4,
+        "budget": 100,
+        "trap": false
+      },
       "split_path": "HF_DATA/mealy/dedeucerl_mealy_n4_b100_test.json",
       "hf_jsonl": "HF_DATA/mealy/hf_mealy_n4_b100_test.jsonl"
     },
+    {
+      "skin": "mealy",
+      "subset": "train",
+      "seeds": "100-10099",
+      "params": {
+        "n_states": 4,
+        "budget": 100,
+        "trap": false
+      },
+      "split_path": "HF_DATA/mealy/dedeucerl_mealy_n4_b100_train.json",
+      "hf_jsonl": "HF_DATA/mealy/hf_mealy_n4_b100_train.jsonl"
+    },
     {
       "skin": "protocol",
       "subset": "test",
+      "seeds": "0-99",
+      "params": {
+        "n_endpoints": 5,
+        "n_states": 4,
+        "budget": 120,
+        "trap": false
+      },
       "split_path": "HF_DATA/protocol/dedeucerl_protocol_n4_e5_b120_test.json",
       "hf_jsonl": "HF_DATA/protocol/hf_protocol_n4_e5_b120_test.jsonl"
     },
+    {
+      "skin": "protocol",
+      "subset": "train",
+      "seeds": "100-10099",
+      "params": {
+        "n_endpoints": 5,
+        "n_states": 4,
+        "budget": 120,
+        "trap": false
+      },
+      "split_path": "HF_DATA/protocol/dedeucerl_protocol_n4_e5_b120_train.json",
+      "hf_jsonl": "HF_DATA/protocol/hf_protocol_n4_e5_b120_train.jsonl"
+    },
     {
       "skin": "apienv",
       "subset": "test",
+      "seeds": "0-99",
+      "params": {
+        "n_endpoints": 7,
+        "n_states": 7,
+        "budget": 200,
+        "trap": false
+      },
       "split_path": "HF_DATA/apienv/dedeucerl_apienv_n7_e7_b200_test.json",
       "hf_jsonl": "HF_DATA/apienv/hf_apienv_n7_e7_b200_test.jsonl"
+    },
+    {
+      "skin": "apienv",
+      "subset": "train",
+      "seeds": "100-10099",
+      "params": {
+        "n_endpoints": 7,
+        "n_states": 7,
+        "budget": 200,
+        "trap": false
+      },
+      "split_path": "HF_DATA/apienv/dedeucerl_apienv_n7_e7_b200_train.json",
+      "hf_jsonl": "HF_DATA/apienv/hf_apienv_n7_e7_b200_train.jsonl"
+    },
+    {
+      "skin": "exprpolicy",
+      "subset": "test",
+      "seeds": "0-99",
+      "params": {
+        "budget": 60,
+        "n_public": 8,
+        "n_hidden": 80,
+        "trap": false
+      },
+      "split_path": "HF_DATA/exprpolicy/dedeucerl_exprpolicy_b60_test.json",
+      "hf_jsonl": "HF_DATA/exprpolicy/hf_exprpolicy_b60_test.jsonl"
+    },
+    {
+      "skin": "exprpolicy",
+      "subset": "train",
+      "seeds": "100-10099",
+      "params": {
+        "budget": 60,
+        "n_public": 8,
+        "n_hidden": 80,
+        "trap": false
+      },
+      "split_path": "HF_DATA/exprpolicy/dedeucerl_exprpolicy_b60_train.json",
+      "hf_jsonl": "HF_DATA/exprpolicy/hf_exprpolicy_b60_train.jsonl"
     }
   ]
 }

README.md

@@ -7,67 +7,112 @@ tags:
 - tool-calling
 - synthetic
 - verifiers
+- agents
+- benchmark
 configs:
 - config_name: mealy_n4_b100
   data_files:
   - split: test
     path: "mealy/hf_mealy_n4_b100_test.jsonl"
+  - split: train
+    path: "mealy/hf_mealy_n4_b100_train.jsonl"
 - config_name: protocol_n4_e5_b120
   data_files:
   - split: test
     path: "protocol/hf_protocol_n4_e5_b120_test.jsonl"
+  - split: train
+    path: "protocol/hf_protocol_n4_e5_b120_train.jsonl"
 - config_name: apienv_n7_e7_b200
   data_files:
   - split: test
     path: "apienv/hf_apienv_n7_e7_b200_test.jsonl"
+  - split: train
+    path: "apienv/hf_apienv_n7_e7_b200_train.jsonl"
+- config_name: exprpolicy_b60
+  data_files:
+  - split: test
+    path: "exprpolicy/hf_exprpolicy_b60_test.jsonl"
+  - split: train
+    path: "exprpolicy/hf_exprpolicy_b60_train.jsonl"
 ---
 
 # DedeuceRL Splits
 
-This dataset provides public, reproducible task splits for the DedeuceRL benchmark.
+Task splits for the DedeuceRL benchmark — a modular framework for active system identification and code debugging tasks.
 
-- Source code: https://github.com/AashVed/DedeuceRL
+- Source: https://github.com/AashVed/DedeuceRL
 - License: MIT
 
-The splits are intended for both model evaluation and RL training (the environment consumes the `answer` field to configure the hidden system; agents should only see the `prompt`).
+## Skins
+
+| Skin | Task | Description |
+|------|------|-------------|
+| **Mealy** | System Identification | Identify hidden Mealy machine (finite-state transducer) |
+| **Protocol** | API Reverse Engineering | Discover state-dependent REST API behavior |
+| **APIEnv** | SaaS API Identification | Realistic stateful SaaS API with variants |
+| **ExprPolicy** | Code Debugging | Fix buggy typed policy DSL expression |
 
-## Contents
+## Splits Summary
 
-Each environment provides two file formats:
+| Skin | Config | Test | Train |
+|------|--------|------|-------|
+| Mealy | n_states=4, budget=100 | 100 (0-99) | 10,000 (100-10099) |
+| Protocol | n_endpoints=5, n_states=4, budget=120 | 100 (0-99) | 10,000 (100-10099) |
+| APIEnv | n_endpoints=7, n_states=7, budget=200 | 100 (0-99) | 10,000 (100-10099) |
+| ExprPolicy | budget=60, n_public=8, n_hidden=80 | 100 (0-99) | 10,000 (100-10099) |
 
-- `dedeucerl_*.json`: Runner-friendly DedeuceRL split format.
-  - Use directly with `dedeucerl-eval --split ...`.
-- `hf_*.jsonl`: Hugging Face friendly JSONL.
-  - One episode per line with keys: `skin`, `subset`, `seed`, `prompt`, `answer`.
-  - Compatible with `datasets.load_dataset()` and the HF dataset viewer.
+All splits generated with `trap=False`.
 
-## Load with Hugging Face Datasets
+## Usage
+
+### Load with HuggingFace Datasets
 
 ```python
 from datasets import load_dataset
 
-ds = load_dataset("comfortably-dumb/DedeuceRL", "mealy_n4_b100", split="test")
-print(ds[0].keys())
-```
+# Load Mealy training set
+ds = load_dataset("comfortably-dumb/DedeuceRL", "mealy_n4_b100", split="train")
 
-## Run with DedeuceRL
+# Load ExprPolicy test set
+ds_test = load_dataset("comfortably-dumb/DedeuceRL", "exprpolicy_b60", split="test")
+```
 
-Download any `dedeucerl_*.json` file and run:
+### Run with DedeuceRL CLI
 
 ```bash
 dedeucerl-eval \
   --skin mealy \
   --split mealy/dedeucerl_mealy_n4_b100_test.json \
-  --model heuristic:none \
+  --model openai:gpt-4o \
   --out results.jsonl
 ```
 
-## Re-generate
-
-These files were generated with `dedeucerl-generate` using:
-
-- Mealy: `n_states=4`, `budget=100`, `trap=false`, seeds `0..99`
-- Protocol: `n_endpoints=5`, `n_states=4`, `budget=120`, `trap=false`, seeds `0..99`
-- APIEnv: `n_endpoints=7`, `n_states=7`, `budget=200`, `trap=false`, seeds `0..99`
-
-See `GENERATION.json` for a machine-readable summary.
+## Schema
+
+Each episode contains:
+- `skin`: Environment type
+- `subset`: Split name (`train`, `test`)
+- `seed`: Reproducibility seed
+- `prompt`: System + user messages for the agent
+- `answer`: JSON-encoded ground truth
+
+## Generation Parameters
+
+| Skin | Parameters |
+|------|------------|
+| Mealy | `n_states=4`, `budget=100`, `trap=false` |
+| Protocol | `n_endpoints=5`, `n_states=4`, `budget=120`, `trap=false` |
+| APIEnv | `n_endpoints=7`, `n_states=7`, `budget=200`, `trap=false` |
+| ExprPolicy | `budget=60`, `n_public=8`, `n_hidden=80`, `trap=false` |
+
+## Citation
+
+```bibtex
+@software{dedeucerl2025,
+  title = {DedeuceRL: A Modular Framework for Active System Identification Benchmarks},
+  author = {Vedansh},
+  year = {2025},
+  url = {https://github.com/AashVed/DedeuceRL},
+  doi = {10.5281/zenodo.18280315}
+}
+```

apienv/dedeucerl_apienv_n7_e7_b200_train.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0de3cb92ba6929e5efe8f15910af86637f91569493bd7a89639dc2450c4608d
+size 209620458

apienv/hf_apienv_n7_e7_b200_train.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b2365323f7c0e1f4fbb2c10320bd6930fa29ad89de97464ca10329b25925099
+size 60620339

exprpolicy/dedeucerl_exprpolicy_b60_train.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c7888b7e8c01443a8018059d636e13a5fa086bdef91be4776976411fc2a22d28
+size 265390771

exprpolicy/hf_exprpolicy_b60_train.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e9404661e189ecaf8e4005a5f0d875b7d67de7cbe5c7c03c433bc659f08afe30
+size 155364310

mealy/dedeucerl_mealy_n4_b100_train.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:82435ee000e619a9469f4ba67bae944ef2329ea7e17471fceb40628e8f68def0
+size 13539297

protocol/dedeucerl_protocol_n4_e5_b120_train.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da286ad38b6543712389eab6209882e9be64a093bd274f8b29716725cda116ce
+size 51279199

protocol/hf_protocol_n4_e5_b120_train.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7d95e1ac9ef1410793e152c403efea097ded40365ce1a8a0117c2a3a0b9ceb9c
+size 12827416