webxos commited on
Commit
a6a00f6
·
verified ·
1 Parent(s): 62a500e

Upload 5 files

Browse files
Files changed (5) hide show
  1. README.md +82 -0
  2. data/test.jsonl +0 -0
  3. data/train.jsonl +0 -0
  4. data/validation.jsonl +0 -0
  5. dataset_card.json +132 -0
README.md ADDED
@@ -0,0 +1,82 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # OPENCHEF!: AGENTIC v2.0 - Multi-Agent Routing Dataset
2
+
3
+ ## Dataset Description
4
+
5
+ This dataset contains simulated multi-agent sessions generated by OPENCHEF!: AGENTIC v2.0, designed for training AI agents in tool use and routing decisions.
6
+
7
+ ### Dataset Summary
8
+ - **Total Sessions**: 770
9
+ - **Train**: 539 (70.0%)
10
+ - **Validation**: 127 (16.5%)
11
+ - **Test**: 104 (13.5%)
12
+ - **Success Rate**: 92.9%
13
+ - **Generated**: 2026-01-30T10:54:18.423Z
14
+
15
+ ### Supported Tasks
16
+ - Tool calling with function-calling format
17
+ - Multi-agent routing decisions
18
+ - Error recovery from routing mistakes
19
+ - Agent coordination
20
+
21
+ ### Data Fields
22
+
23
+ Each example contains:
24
+ - `id`: Unique session identifier
25
+ - `split`: Dataset split (train/validation/test)
26
+ - `task_description`: User instruction for the session
27
+ - `agents_involved`: List of agent types in the session
28
+ - `full_trace`: Complete session trace with thoughts, actions, and observations
29
+ - `final_success`: Whether the session completed successfully
30
+ - `steps_count`: Number of turns in the session
31
+ - `tool_use_count`: Number of tool calls made
32
+ - `routing_decisions`: Number of routing decisions made
33
+ - `metadata`: Generation metadata
34
+
35
+ ### Data Splits
36
+
37
+ The dataset is split into train (70%), validation (15%), and test (15%) sets.
38
+
39
+ ## Usage
40
+
41
+ ### For Hugging Face Datasets
42
+ ```python
43
+ from datasets import load_dataset
44
+
45
+ # Load directly from JSONL files
46
+ dataset = load_dataset('json', data_files={
47
+ 'train': 'data/train.jsonl',
48
+ 'validation': 'data/validation.jsonl',
49
+ 'test': 'data/test.jsonl'
50
+ })
51
+ ```
52
+
53
+ ### For Training Agentic Models
54
+ This dataset is optimized for training models on:
55
+ - Tool calling behavior
56
+ - Routing decisions between specialized agents
57
+ - Error recovery from incorrect routing
58
+ - Multi-agent coordination
59
+
60
+ ## Generation Details
61
+
62
+ - **Generator**: OPENCHEF!: AGENTIC v2.0
63
+ - **Error Rate**: 15% intentional routing errors
64
+ - **Agent Types**: Router, Email Agent, Calendar Agent, Code Agent, Web Agent, User Proxy, System
65
+ - **Tool Types**: Email search/send, Calendar management, Code execution, Web search, File operations
66
+ - **Format**: Function-calling with ReAct-style traces
67
+
68
+ ## Citation
69
+
70
+ If you use this dataset, please cite:
71
+ ```
72
+ @software{openchef2024,
73
+ title = {OPENCHEF!: AGENTIC v2.0 - Multi-Agent Routing Dataset Generator},
74
+ author = {OPENCLAW.ai Team},
75
+ year = {2024},
76
+ url = {https://github.com/openclaw/openchef}
77
+ }
78
+ ```
79
+
80
+ ## License
81
+
82
+ MIT License
data/test.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
data/train.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
data/validation.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
dataset_card.json ADDED
@@ -0,0 +1,132 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "description": "Multi-agent routing and tool use dataset for training AI agents",
3
+ "citation": "OPENCHEF!: AGENTIC v2.0 - Multi-Agent Routing Dataset",
4
+ "homepage": "https://github.com/openclaw/openchef",
5
+ "license": "mit",
6
+ "features": {
7
+ "id": {
8
+ "dtype": "string",
9
+ "id": null,
10
+ "_type": "Value"
11
+ },
12
+ "split": {
13
+ "dtype": "string",
14
+ "id": null,
15
+ "_type": "Value"
16
+ },
17
+ "task_description": {
18
+ "dtype": "string",
19
+ "id": null,
20
+ "_type": "Value"
21
+ },
22
+ "agents_involved": {
23
+ "feature": {
24
+ "dtype": "string",
25
+ "id": null,
26
+ "_type": "Value"
27
+ },
28
+ "length": -1,
29
+ "id": null,
30
+ "_type": "Sequence"
31
+ },
32
+ "full_trace": {
33
+ "feature": {
34
+ "agent_name": {
35
+ "dtype": "string",
36
+ "id": null,
37
+ "_type": "Value"
38
+ },
39
+ "role": {
40
+ "dtype": "string",
41
+ "id": null,
42
+ "_type": "Value"
43
+ },
44
+ "content": {
45
+ "dtype": "string",
46
+ "id": null,
47
+ "_type": "Value"
48
+ },
49
+ "tool_call": {
50
+ "dtype": "string",
51
+ "id": null,
52
+ "_type": "Value"
53
+ },
54
+ "tool_result": {
55
+ "dtype": "string",
56
+ "id": null,
57
+ "_type": "Value"
58
+ },
59
+ "timestamp_ms": {
60
+ "dtype": "int64",
61
+ "id": null,
62
+ "_type": "Value"
63
+ }
64
+ },
65
+ "length": -1,
66
+ "id": null,
67
+ "_type": "Sequence"
68
+ },
69
+ "final_success": {
70
+ "dtype": "bool",
71
+ "id": null,
72
+ "_type": "Value"
73
+ },
74
+ "steps_count": {
75
+ "dtype": "int64",
76
+ "id": null,
77
+ "_type": "Value"
78
+ },
79
+ "tool_use_count": {
80
+ "dtype": "int64",
81
+ "id": null,
82
+ "_type": "Value"
83
+ },
84
+ "routing_decisions": {
85
+ "dtype": "int64",
86
+ "id": null,
87
+ "_type": "Value"
88
+ },
89
+ "metadata": {
90
+ "dtype": "string",
91
+ "id": null,
92
+ "_type": "Value"
93
+ }
94
+ },
95
+ "post_processed": null,
96
+ "supervised_keys": null,
97
+ "task_templates": null,
98
+ "builder_name": "openchef_agentic",
99
+ "config_name": "default",
100
+ "version": {
101
+ "version_str": "2.0.0",
102
+ "description": null,
103
+ "major": 2,
104
+ "minor": 0,
105
+ "patch": 0
106
+ },
107
+ "splits": {
108
+ "train": {
109
+ "name": "train",
110
+ "num_bytes": 0,
111
+ "num_examples": 539,
112
+ "dataset_name": "openchef_agentic"
113
+ },
114
+ "validation": {
115
+ "name": "validation",
116
+ "num_bytes": 0,
117
+ "num_examples": 127,
118
+ "dataset_name": "openchef_agentic"
119
+ },
120
+ "test": {
121
+ "name": "test",
122
+ "num_bytes": 0,
123
+ "num_examples": 104,
124
+ "dataset_name": "openchef_agentic"
125
+ }
126
+ },
127
+ "download_checksums": null,
128
+ "download_size": 0,
129
+ "post_processing_size": null,
130
+ "dataset_size": 0,
131
+ "size_in_bytes": 0
132
+ }