Repository navigation
Expand file tree
/
Copy pathworkflow.json
More file actions
155 lines (155 loc) · 6.98 KB
/
Copy pathworkflow.json
File metadata and controls
155 lines (155 loc) · 6.98 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
{
"schemaVersion": 2,
"name": "autoresearch：Git keep/discard 的完整实验循环",
"agents": {
"main": {
"model": "nex-agi/Nex-N2-Pro",
"harness": "codex",
"skills": [
"./skills/scienceevolve-autoresearch"
],
"autoSkills": [],
"mcpServers": [],
"instructions": "检查用户提供的autoresearch Task源码、Git、uv、数据缓存及单GPU环境。遵循Skill建立独立实验分支，初始设置得到确认后先交train跑未修改的基线。随后propose→train→select循环。保存控制状态与Git父版本，用户未停止时持续运行，不设10次之类自创上限，不按无提升自动结束。实验崩溃、超时、候选被淘汰后都继续搜索；严重环境故障交用户处理。收到明确停止指令后保存当前最佳版本与完整results.tsv并结束；用户指定预算时按其覆盖上游持续运行模式并注明。\n\n读取所配Skill中的完整执行协议。Task与research workflow分开；本方法没有Evolver，不调用apply_workflow。缺少数据、工具或权限时交Main明确阻断，不省略阶段或编造结果。产物、原始日志、失败状态必须落盘，恢复时核对已有状态防止重复执行。具体模型/工具可按需求等价替换，不能删除能力、减少上游默认步骤/候选数或擅自改停止规则。",
"permission": "workspace",
"webSearch": true,
"sessionPersistence": true,
"decisionBy": "main",
"gatewayReasoningOption": "high",
"proxy": {
"mode": "direct"
}
},
"propose": {
"model": "nex-agi/Nex-N2-Pro",
"harness": "codex",
"skills": [
"./skills/scienceevolve-autoresearch"
],
"autoSkills": [],
"mcpServers": [],
"instructions": "从当前已接受Git提交提出一个机制/架构/优化器或训练参数实验；只修改train.py，提交改动并记录父commit和候选commit，然后交train。考虑复杂度与收益，阅读历史失败避免重复；不能改prepare.py、评估器、数据划分、时间窗口或依赖。\n\n读取所配Skill中的完整执行协议。Task与research workflow分开；本方法没有Evolver，不调用apply_workflow。缺少数据、工具或权限时交Main明确阻断，不省略阶段或编造结果。产物、原始日志、失败状态必须落盘，恢复时核对已有状态防止重复执行。具体模型/工具可按需求等价替换，不能删除能力、减少上游默认步骤/候选数或擅自改停止规则。",
"permission": "workspace",
"webSearch": false,
"sessionPersistence": false,
"decisionBy": "main",
"gatewayReasoningOption": "high",
"proxy": {
"mode": "direct"
}
},
"train": {
"model": "nex-agi/Nex-N2-Pro",
"harness": "codex",
"skills": [
"./skills/scienceevolve-autoresearch"
],
"autoSkills": [],
"mcpServers": [],
"instructions": "按Skill运行基线或已提交候选的uv run train.py，重定向日志；保留5分钟训练时间窗（不含启动/编译），总墙钟超过10分钟终止该Task进程并记录timeout。保存退出码、val_bpb、peak_vram_mb和完整日志；简单崩溃可在有限几次修复后重跑，每次版本和日志可追溯，结构性失败及时判crash。没有val_bpb不能判成功；交select。\n\n读取所配Skill中的完整执行协议。Task与research workflow分开；本方法没有Evolver，不调用apply_workflow。缺少数据、工具或权限时交Main明确阻断，不省略阶段或编造结果。产物、原始日志、失败状态必须落盘，恢复时核对已有状态防止重复执行。具体模型/工具可按需求等价替换，不能删除能力、减少上游默认步骤/候选数或擅自改停止规则。",
"permission": "workspace",
"webSearch": false,
"sessionPersistence": false,
"decisionBy": "main",
"gatewayReasoningOption": "high",
"proxy": {
"mode": "direct"
}
},
"select": {
"model": "nex-agi/Nex-N2-Pro",
"harness": "codex",
"skills": [
"./skills/scienceevolve-autoresearch"
],
"autoSkills": [],
"mcpServers": [],
"instructions": "按照Skill的指标/复杂度权衡决定keep/discard/crash，写原5列results.tsv并保留历史。正常且更优/同等更简单的候选才可保留；失败记录0是日志占位，不能作为最低val_bpb获胜。discard/crash通过Git恢复本轮parent。只在预检通过的专用干净实验分支操作，不能重置用户无关改动。返回Main继续。\n\n读取所配Skill中的完整执行协议。Task与research workflow分开；本方法没有Evolver，不调用apply_workflow。缺少数据、工具或权限时交Main明确阻断，不省略阶段或编造结果。产物、原始日志、失败状态必须落盘，恢复时核对已有状态防止重复执行。具体模型/工具可按需求等价替换，不能删除能力、减少上游默认步骤/候选数或擅自改停止规则。",
"permission": "workspace",
"webSearch": false,
"sessionPersistence": false,
"decisionBy": "main",
"gatewayReasoningOption": "high",
"proxy": {
"mode": "direct"
}
}
},
"orchestration": {
"agents": [
{
"name": "main",
"template": "main"
},
{
"name": "propose",
"template": "general"
},
{
"name": "train",
"template": "general"
},
{
"name": "select",
"template": "general"
}
],
"input": {
"source": "input",
"target": "main"
},
"routes": {
"main": [
{
"target": "propose",
"kind": "weak"
},
{
"target": "train",
"kind": "weak"
}
],
"propose": [
{
"target": "train",
"kind": "strong"
}
],
"train": [
{
"target": "select",
"kind": "strong"
}
],
"select": [
{
"target": "main",
"kind": "strong"
}
]
},
"decisionPlans": {
"main": "遵循完整协议、真实状态和阶段门槛调度。",
"propose": "按完整协议执行当前职责并交 train。",
"train": "按完整协议执行当前职责并交 select。",
"select": "按完整协议执行当前职责并交 main。"
},
"allowUserIntervention": true,
"allowMainReconfiguration": false,
"termination": {
"agent": "main",
"condition": "原流程规定的终点已达到且交付完整；持续运行方法仅在用户明确要求结束时完成。环境阻断不能当成功。",
"decisionBy": "main",
"automatic": false
},
"resultFormat": {
"presets": [
"可复现证据"
],
"custom": "完整状态、候选/版本、原始结果、分析及终止状态，路径见Skill。"
}
},
"git": {
"enabled": false
}
}