yzzhao commited on
Commit
3b30eee
·
verified ·
1 Parent(s): f150553

Add first matched RoboTwin outcomes with full sessions and policy traces

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitattributes +2 -0
  2. app.js +1 -1
  3. data.json +31 -5
  4. experiments/robotwin-native/kinex-1/artifacts.json +109 -0
  5. experiments/robotwin-native/kinex-1/combined.mp4 +3 -0
  6. experiments/robotwin-native/kinex-1/detail.json +1375 -0
  7. experiments/robotwin-native/kinex-1/effective-model.json +16 -0
  8. experiments/robotwin-native/kinex-1/episode-report.json +497 -0
  9. experiments/robotwin-native/kinex-1/episode.json +15 -0
  10. experiments/robotwin-native/kinex-1/evidence.tar.gz +3 -0
  11. experiments/robotwin-native/kinex-1/instruction.json +14 -0
  12. experiments/robotwin-native/kinex-1/media.json +101 -0
  13. experiments/robotwin-native/kinex-1/observations/01d7a694-1.png +0 -0
  14. experiments/robotwin-native/kinex-1/observations/0bd66e5e-1.png +0 -0
  15. experiments/robotwin-native/kinex-1/observations/0d7d46d8-1.png +0 -0
  16. experiments/robotwin-native/kinex-1/observations/13f789a0-1.png +0 -0
  17. experiments/robotwin-native/kinex-1/observations/1953a6e9-1.png +0 -0
  18. experiments/robotwin-native/kinex-1/observations/245fef43-1.png +0 -0
  19. experiments/robotwin-native/kinex-1/observations/3d9b5793-1.png +0 -0
  20. experiments/robotwin-native/kinex-1/observations/451da7eb-1.png +0 -0
  21. experiments/robotwin-native/kinex-1/observations/4a6cdfeb-1.png +0 -0
  22. experiments/robotwin-native/kinex-1/observations/4f33bd50-1.png +0 -0
  23. experiments/robotwin-native/kinex-1/observations/5c5beb92-1.png +0 -0
  24. experiments/robotwin-native/kinex-1/observations/5df3e384-1.png +0 -0
  25. experiments/robotwin-native/kinex-1/observations/5ffc47a8-1.png +0 -0
  26. experiments/robotwin-native/kinex-1/observations/797caa5e-1.png +0 -0
  27. experiments/robotwin-native/kinex-1/observations/86db988c-1.png +0 -0
  28. experiments/robotwin-native/kinex-1/observations/86db988c-2.png +0 -0
  29. experiments/robotwin-native/kinex-1/observations/a51b5070-1.png +0 -0
  30. experiments/robotwin-native/kinex-1/observations/a8adce0d-1.png +0 -0
  31. experiments/robotwin-native/kinex-1/observations/af212c1a-1.png +0 -0
  32. experiments/robotwin-native/kinex-1/observations/b5d4c986-1.png +0 -0
  33. experiments/robotwin-native/kinex-1/observations/b833ee1b-1.png +0 -0
  34. experiments/robotwin-native/kinex-1/observations/bab02154-1.png +0 -0
  35. experiments/robotwin-native/kinex-1/observations/bf1286a6-1.png +0 -0
  36. experiments/robotwin-native/kinex-1/observations/c1d037d5-1.png +0 -0
  37. experiments/robotwin-native/kinex-1/observations/c3226fa8-1.png +0 -0
  38. experiments/robotwin-native/kinex-1/observations/db1b4a52-1.png +0 -0
  39. experiments/robotwin-native/kinex-1/observations/e7cd2d3f-1.png +0 -0
  40. experiments/robotwin-native/kinex-1/observations/e84d68b9-1.png +0 -0
  41. experiments/robotwin-native/kinex-1/observations/f0765732-1.png +0 -0
  42. experiments/robotwin-native/kinex-1/observations/f9d2862d-1.png +0 -0
  43. experiments/robotwin-native/kinex-1/policy-calls.json +1 -0
  44. experiments/robotwin-native/kinex-1/poster.jpg +0 -0
  45. experiments/robotwin-native/kinex-1/provider-requests.jsonl +0 -0
  46. experiments/robotwin-native/kinex-1/session.json +0 -0
  47. experiments/robotwin-native/kinex-1/session.jsonl +0 -0
  48. experiments/robotwin-native/kinex-1/terminal.jpg +0 -0
  49. experiments/robotwin-native/kinex-1/tool-calls.json +0 -0
  50. experiments/robotwin-native/kinex-1/trace.json +0 -0
.gitattributes CHANGED
@@ -54,3 +54,5 @@ data/b2/video.mp4 filter=lfs diff=lfs merge=lfs -text
54
  probes/bridge-20261028/poster.jpg filter=lfs diff=lfs merge=lfs -text
55
  probes/bridge-20261028/terminal.jpg filter=lfs diff=lfs merge=lfs -text
56
  probes/bridge-20261028/video.mp4 filter=lfs diff=lfs merge=lfs -text
 
 
 
54
  probes/bridge-20261028/poster.jpg filter=lfs diff=lfs merge=lfs -text
55
  probes/bridge-20261028/terminal.jpg filter=lfs diff=lfs merge=lfs -text
56
  probes/bridge-20261028/video.mp4 filter=lfs diff=lfs merge=lfs -text
57
+ experiments/robotwin-native/standalone-1/combined.mp4 filter=lfs diff=lfs merge=lfs -text
58
+ experiments/robotwin-native/kinex-1/combined.mp4 filter=lfs diff=lfs merge=lfs -text
app.js CHANGED
@@ -14,7 +14,7 @@ const metrics=`<div class="metrics">${[['Native steps',num(e.control_steps)],['T
14
  if(tab==='overview')target.innerHTML=`<div class="panel">${e.video?`<video id="episode-video" class="video" controls playsinline preload="metadata" poster="${esc(e.poster||'')}"><source src="${esc(e.video)}" type="video/mp4"></video>`:'<div class="empty">Recording is not available yet</div>'}${metrics}<p class="note">${esc(d.cost_note||'API-equivalent cost is an estimate, not a subscription charge.')} ${esc(d.media_note||'')}</p></div><div class="panel"><h2>Observed behavior</h2><p>${esc(d.observation||'Analysis pending.')}</p><a href="${e.path}/detail.json">Episode record ↗</a></div>`;
15
  else if(tab==='instruction')target.innerHTML=`<div class="panel"><h2>Task instruction</h2><div class="text">${esc(d.instruction)}</div></div>`;
16
  else if(tab==='artifacts')target.innerHTML=`<div class="panel files"><h2>Recorded artifacts</h2>${(d.artifacts||[]).map(a=>`<a href="${esc(a.path)}" target="_blank" rel="noopener">${esc(a.name)} <span class="muted">${num(a.bytes)} bytes ↗</span></a>`).join('')}</div>`;
17
- else if(tab==='session'){const v=await load(e.path+'/session.json');target.innerHTML=`<p class="muted">${num(v.length)} messages · <a href="${e.path}/session.json">Download full session</a></p>${v.map(m=>`<article class="message"><b>${esc(m.role)} ${esc(m.name||'')}</b><div class="text">${esc(m.text)}</div>${m.data?`<details><summary>Recorded payload</summary><pre>${esc(JSON.stringify(m.data,null,2))}</pre></details>`:''}</article>`).join('')||'<div class="empty">Standalone evaluation has no planning-agent session.</div>'}`;}
18
  else if(tab==='tools'){const v=await load(e.path+'/tool-calls.json');target.innerHTML=`<p class="muted">${num(v.length)} calls · <a href="${e.path}/tool-calls.json">Download tool calls</a></p>${v.map((m,i)=>`<details><summary>${i+1}. ${esc(m.name)} <span class="muted">${esc(m.summary||'')}</span></summary><h3>Input</h3><pre>${esc(JSON.stringify(m.input,null,2))}</pre><h3>Output</h3><pre>${esc(typeof m.output==='string'?m.output:JSON.stringify(m.output,null,2))}</pre></details>`).join('')||'<div class="empty">No agent tool calls</div>'}`;}
19
  else if(tab==='policy'){const v=await load(e.path+'/policy-calls.json');target.innerHTML=`<p class="muted">${num(v.length)} inferences · <a href="${e.path}/policy-calls.json">Download all input / output records</a></p>${v.map((m,i)=>`<details><summary>${i+1}. ${esc(m.task)} <span class="muted">${m.inference_s?.toFixed(2)||'—'} s</span></summary><div class="images">${(m.images||[]).map(x=>`<a href="${esc(x)}" target="_blank"><img loading="lazy" src="${esc(x)}" alt="Policy input observation"></a>`).join('')}</div><h3>Model output and submitted interface</h3><pre>${esc(JSON.stringify(m.output,null,2))}</pre></details>`).join('')||'<div class="empty">No VLA inference recorded</div>'}`;}}
20
  function protocol(){$('#main').innerHTML=`<div class="eyebrow">Experiment protocol</div><h1>One task. Three control choices.</h1><div class="panel"><h2>Execution</h2><table><tbody>${[['Suite','RLE-Bench task03 · RoboTwin 2.0 · ALOHA AgileX'],['VLA','StarVLA/Qwen3-VL-OFT-Robotwin2 · steps 40,000'],['Agent','Kinex · GPT-6 Astra · high'],['Policy interface','Native run_policy · one inference per call · agent owns the outer loop'],['Control','25 Hz native joint / gripper frames; original task-specific limits'],['Selection','First three task03 tasks, seed 0; frozen before outcome inspection'],['Scope','Frozen-policy behavior only; no encoder or progress monitor']].map(([k,v])=>`<tr><th>${k}</th><td>${v}</td></tr>`).join('')}</tbody></table></div><p class="muted">This fixed-rate device protocol differs from upstream VLA trajectory execution. Published StarVLA benchmark scores are not a direct comparison. RoboTwin 2.0-Plus is a separate suite and is not included here.</p><a href="sources.json">Source and model identities ↗</a>`}
 
14
  if(tab==='overview')target.innerHTML=`<div class="panel">${e.video?`<video id="episode-video" class="video" controls playsinline preload="metadata" poster="${esc(e.poster||'')}"><source src="${esc(e.video)}" type="video/mp4"></video>`:'<div class="empty">Recording is not available yet</div>'}${metrics}<p class="note">${esc(d.cost_note||'API-equivalent cost is an estimate, not a subscription charge.')} ${esc(d.media_note||'')}</p></div><div class="panel"><h2>Observed behavior</h2><p>${esc(d.observation||'Analysis pending.')}</p><a href="${e.path}/detail.json">Episode record ↗</a></div>`;
15
  else if(tab==='instruction')target.innerHTML=`<div class="panel"><h2>Task instruction</h2><div class="text">${esc(d.instruction)}</div></div>`;
16
  else if(tab==='artifacts')target.innerHTML=`<div class="panel files"><h2>Recorded artifacts</h2>${(d.artifacts||[]).map(a=>`<a href="${esc(a.path)}" target="_blank" rel="noopener">${esc(a.name)} <span class="muted">${num(a.bytes)} bytes ↗</span></a>`).join('')}</div>`;
17
+ else if(tab==='session'){const v=await load(e.path+'/session.json');target.innerHTML=`<p class="muted">${num(v.length)} messages · <a href="${e.path}/session.json">Download full session</a></p>${v.map(m=>`<article class="message"><b>${esc(m.role)} ${esc(m.name||'')}</b><div class="text">${esc(m.text)}</div><div class="images">${(m.images||[]).map(x=>`<a href="${esc(x)}" target="_blank"><img loading="lazy" src="${esc(x)}" alt="Agent observation"></a>`).join('')}</div>${m.data?`<details><summary>Recorded payload</summary><pre>${esc(JSON.stringify(m.data,null,2))}</pre></details>`:''}</article>`).join('')||'<div class="empty">Standalone evaluation has no planning-agent session.</div>'}`;}
18
  else if(tab==='tools'){const v=await load(e.path+'/tool-calls.json');target.innerHTML=`<p class="muted">${num(v.length)} calls · <a href="${e.path}/tool-calls.json">Download tool calls</a></p>${v.map((m,i)=>`<details><summary>${i+1}. ${esc(m.name)} <span class="muted">${esc(m.summary||'')}</span></summary><h3>Input</h3><pre>${esc(JSON.stringify(m.input,null,2))}</pre><h3>Output</h3><pre>${esc(typeof m.output==='string'?m.output:JSON.stringify(m.output,null,2))}</pre></details>`).join('')||'<div class="empty">No agent tool calls</div>'}`;}
19
  else if(tab==='policy'){const v=await load(e.path+'/policy-calls.json');target.innerHTML=`<p class="muted">${num(v.length)} inferences · <a href="${e.path}/policy-calls.json">Download all input / output records</a></p>${v.map((m,i)=>`<details><summary>${i+1}. ${esc(m.task)} <span class="muted">${m.inference_s?.toFixed(2)||'—'} s</span></summary><div class="images">${(m.images||[]).map(x=>`<a href="${esc(x)}" target="_blank"><img loading="lazy" src="${esc(x)}" alt="Policy input observation"></a>`).join('')}</div><h3>Model output and submitted interface</h3><pre>${esc(JSON.stringify(m.output,null,2))}</pre></details>`).join('')||'<div class="empty">No VLA inference recorded</div>'}`;}}
20
  function protocol(){$('#main').innerHTML=`<div class="eyebrow">Experiment protocol</div><h1>One task. Three control choices.</h1><div class="panel"><h2>Execution</h2><table><tbody>${[['Suite','RLE-Bench task03 · RoboTwin 2.0 · ALOHA AgileX'],['VLA','StarVLA/Qwen3-VL-OFT-Robotwin2 · steps 40,000'],['Agent','Kinex · GPT-6 Astra · high'],['Policy interface','Native run_policy · one inference per call · agent owns the outer loop'],['Control','25 Hz native joint / gripper frames; original task-specific limits'],['Selection','First three task03 tasks, seed 0; frozen before outcome inspection'],['Scope','Frozen-policy behavior only; no encoder or progress monitor']].map(([k,v])=>`<tr><th>${k}</th><td>${v}</td></tr>`).join('')}</tbody></table></div><p class="muted">This fixed-rate device protocol differs from upstream VLA trajectory execution. Published StarVLA benchmark scores are not a direct comparison. RoboTwin 2.0-Plus is a separate suite and is not included here.</p><a href="sources.json">Source and model identities ↗</a>`}
data.json CHANGED
@@ -26,7 +26,7 @@
26
  {
27
  "id": "standalone",
28
  "label": "Standalone VLA",
29
- "model": "Qwen3-VL-OFT \u00b7 40k"
30
  },
31
  {
32
  "id": "kinex",
@@ -36,7 +36,7 @@
36
  {
37
  "id": "kinex_policy",
38
  "label": "Kinex + VLA",
39
- "model": "GPT-6 Astra / high \u00b7 native run_policy"
40
  }
41
  ],
42
  "episodes": [
@@ -44,13 +44,39 @@
44
  "id": "standalone-1",
45
  "task": "1",
46
  "condition": "standalone",
47
- "status": "running"
 
 
 
 
 
 
 
 
 
 
 
 
 
48
  },
49
  {
50
  "id": "kinex-1",
51
  "task": "1",
52
  "condition": "kinex",
53
- "status": "running"
 
 
 
 
 
 
 
 
 
 
 
 
 
54
  },
55
  {
56
  "id": "kinex_policy-1",
@@ -95,4 +121,4 @@
95
  "status": "pending"
96
  }
97
  ]
98
- }
 
26
  {
27
  "id": "standalone",
28
  "label": "Standalone VLA",
29
+ "model": "Qwen3-VL-OFT · 40k"
30
  },
31
  {
32
  "id": "kinex",
 
36
  {
37
  "id": "kinex_policy",
38
  "label": "Kinex + VLA",
39
+ "model": "GPT-6 Astra / high · native run_policy"
40
  }
41
  ],
42
  "episodes": [
 
44
  "id": "standalone-1",
45
  "task": "1",
46
  "condition": "standalone",
47
+ "status": "complete",
48
+ "success": false,
49
+ "valid": true,
50
+ "control_steps": 800,
51
+ "tool_calls": 0,
52
+ "vla_calls": 50,
53
+ "input_tokens": 0,
54
+ "output_tokens": 0,
55
+ "cached_input_tokens": 0,
56
+ "api_cost_usd": 0,
57
+ "path": "experiments/robotwin-native/standalone-1",
58
+ "video": "experiments/robotwin-native/standalone-1/combined.mp4",
59
+ "poster": "experiments/robotwin-native/standalone-1/poster.jpg",
60
+ "wall_time_s": 261.565797
61
  },
62
  {
63
  "id": "kinex-1",
64
  "task": "1",
65
  "condition": "kinex",
66
+ "status": "complete",
67
+ "success": true,
68
+ "valid": true,
69
+ "control_steps": 621,
70
+ "tool_calls": 61,
71
+ "vla_calls": 0,
72
+ "input_tokens": 1774043,
73
+ "output_tokens": 8679,
74
+ "cached_input_tokens": 1715200,
75
+ "api_cost_usd": 2.73758,
76
+ "path": "experiments/robotwin-native/kinex-1",
77
+ "video": "experiments/robotwin-native/kinex-1/combined.mp4",
78
+ "poster": "experiments/robotwin-native/kinex-1/poster.jpg",
79
+ "wall_time_s": 635.897416
80
  },
81
  {
82
  "id": "kinex_policy-1",
 
121
  "status": "pending"
122
  }
123
  ]
124
+ }
experiments/robotwin-native/kinex-1/artifacts.json ADDED
@@ -0,0 +1,109 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "evidence": {
3
+ "complete": true,
4
+ "path": "evidence.tar.gz"
5
+ },
6
+ "recording": {
7
+ "complete": true,
8
+ "enabled": true,
9
+ "encoder_queue_drops": 0,
10
+ "lossless": false,
11
+ "manifest": {
12
+ "bag_messages": 0,
13
+ "error": null,
14
+ "fps": 10.0,
15
+ "run_id": "7b75d3e8384a",
16
+ "schema": "rosbridge/recording/1",
17
+ "spectator_views": [
18
+ "third_person",
19
+ "left_wrist",
20
+ "right_wrist"
21
+ ],
22
+ "state": "complete",
23
+ "task": "robotwin/handover-block-aloha-agilex",
24
+ "topics": [],
25
+ "type": "mp4",
26
+ "views": {
27
+ "left_wrist": {
28
+ "dropped": 0,
29
+ "error": null,
30
+ "received": 250,
31
+ "segments": [
32
+ {
33
+ "file": "left_wrist.mp4",
34
+ "first_stamp_ns": 0,
35
+ "fps": 10.0,
36
+ "frames": 250,
37
+ "height": 720,
38
+ "last_stamp_ns": 24840000000,
39
+ "width": 960
40
+ }
41
+ ]
42
+ },
43
+ "right_wrist": {
44
+ "dropped": 0,
45
+ "error": null,
46
+ "received": 250,
47
+ "segments": [
48
+ {
49
+ "file": "right_wrist.mp4",
50
+ "first_stamp_ns": 0,
51
+ "fps": 10.0,
52
+ "frames": 250,
53
+ "height": 720,
54
+ "last_stamp_ns": 24840000000,
55
+ "width": 960
56
+ }
57
+ ]
58
+ },
59
+ "third_person": {
60
+ "dropped": 0,
61
+ "error": null,
62
+ "received": 250,
63
+ "segments": [
64
+ {
65
+ "file": "third_person.mp4",
66
+ "first_stamp_ns": 0,
67
+ "fps": 10.0,
68
+ "frames": 250,
69
+ "height": 720,
70
+ "last_stamp_ns": 24840000000,
71
+ "width": 960
72
+ }
73
+ ]
74
+ }
75
+ }
76
+ },
77
+ "video_available": true,
78
+ "videos": [
79
+ {
80
+ "bytes": 2505135,
81
+ "channel": "spectator",
82
+ "fps": 10.0,
83
+ "frames": 250,
84
+ "path": "recording/third_person.mp4"
85
+ },
86
+ {
87
+ "bytes": 1719860,
88
+ "channel": "spectator",
89
+ "fps": 10.0,
90
+ "frames": 250,
91
+ "path": "recording/left_wrist.mp4"
92
+ },
93
+ {
94
+ "bytes": 1454029,
95
+ "channel": "spectator",
96
+ "fps": 10.0,
97
+ "frames": 250,
98
+ "path": "recording/right_wrist.mp4"
99
+ }
100
+ ]
101
+ },
102
+ "schema": "rlebench/episode-artifacts/1",
103
+ "terminal_evidence": {
104
+ "available": false,
105
+ "complete": false,
106
+ "error": "FileNotFoundError: [Errno 2] No such file or directory: '/run/rlebench/private/episode_01/evidence/terminal.json'"
107
+ },
108
+ "workspace_complete": true
109
+ }
experiments/robotwin-native/kinex-1/combined.mp4 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2c086a9ec449338ab3056eff3315773b24fa89b3737c81a31337ba11a74ed81f
3
+ size 4426774
experiments/robotwin-native/kinex-1/detail.json ADDED
@@ -0,0 +1,1375 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "instruction": "use the left arm to grasp the red block on the table, hand it over to the right arm, and set it down in the center of the blue pad. Release it with the right gripper open",
3
+ "observation": "Kinex completed the handover using its native robot tools and an agent-written controller. The native evaluator reported success at action frame 621; the agent stopped when the world ended during the final release.",
4
+ "usage": {
5
+ "n_input_tokens": 1774043,
6
+ "n_output_tokens": 8679,
7
+ "n_cache_tokens": 1715200,
8
+ "cost_usd": null,
9
+ "n_fresh_input_tokens": 58843,
10
+ "n_cache_read_tokens": 1715200,
11
+ "n_cache_write_tokens": 0
12
+ },
13
+ "usage_details": {
14
+ "accounting": "reported-responses",
15
+ "audit_complete": true,
16
+ "audit_source": "Provider terminal response usage",
17
+ "cache_hit_rate": 0.9668311309252369,
18
+ "cache_read_input_tokens": 1715200,
19
+ "cache_reported_input_tokens": 1774043,
20
+ "cache_write_basis": "provider-reported",
21
+ "cache_write_input_tokens": 0,
22
+ "cache_write_reported_input_tokens": 1774043,
23
+ "cached_input_tokens": 1715200,
24
+ "cost_usd": null,
25
+ "fresh_input_tokens": 58843,
26
+ "harbor_mismatches": {},
27
+ "input_breakdown_complete": true,
28
+ "input_tokens": 1774043,
29
+ "known_cache_write_input_tokens": 0,
30
+ "known_cached_input_tokens": 1715200,
31
+ "known_fresh_input_tokens": 58843,
32
+ "known_input_tokens": 1774043,
33
+ "known_output_tokens": 8679,
34
+ "known_reasoning_output_tokens": 2627,
35
+ "output_tokens": 8679,
36
+ "parser": "harbor/0.22.0",
37
+ "reasoning_output_tokens": 2627,
38
+ "reasoning_reported_output_tokens": 8679,
39
+ "rejected_attempts": 0,
40
+ "reported_responses": {
41
+ "cache_reported_input_tokens": 54,
42
+ "cache_write_input_tokens": 54,
43
+ "cache_write_reported_input_tokens": 54,
44
+ "cached_input_tokens": 54,
45
+ "fresh_input_tokens": 54,
46
+ "input_tokens": 54,
47
+ "output_tokens": 54,
48
+ "reasoning_output_tokens": 54,
49
+ "reasoning_reported_output_tokens": 54
50
+ },
51
+ "request_attempts": 54,
52
+ "response_count": 54,
53
+ "response_ids_complete": true,
54
+ "schema": "rlebench/token-usage/1",
55
+ "source": "Harbor session metrics",
56
+ "total_tokens": 1782722,
57
+ "uncached_input_tokens": 58843,
58
+ "unresolved_attempts": [],
59
+ "usage_records": [
60
+ {
61
+ "cache_write_input_tokens": 0,
62
+ "cached_input_tokens": 0,
63
+ "input_tokens": 11322,
64
+ "output_tokens": 19,
65
+ "reasoning_output_tokens": 0
66
+ },
67
+ {
68
+ "cache_write_input_tokens": 0,
69
+ "cached_input_tokens": 11136,
70
+ "input_tokens": 11362,
71
+ "output_tokens": 29,
72
+ "reasoning_output_tokens": 0
73
+ },
74
+ {
75
+ "cache_write_input_tokens": 0,
76
+ "cached_input_tokens": 0,
77
+ "input_tokens": 11495,
78
+ "output_tokens": 30,
79
+ "reasoning_output_tokens": 0
80
+ },
81
+ {
82
+ "cache_write_input_tokens": 0,
83
+ "cached_input_tokens": 11392,
84
+ "input_tokens": 14101,
85
+ "output_tokens": 223,
86
+ "reasoning_output_tokens": 0
87
+ },
88
+ {
89
+ "cache_write_input_tokens": 0,
90
+ "cached_input_tokens": 14208,
91
+ "input_tokens": 18793,
92
+ "output_tokens": 55,
93
+ "reasoning_output_tokens": 0
94
+ },
95
+ {
96
+ "cache_write_input_tokens": 0,
97
+ "cached_input_tokens": 18688,
98
+ "input_tokens": 20034,
99
+ "output_tokens": 78,
100
+ "reasoning_output_tokens": 0
101
+ },
102
+ {
103
+ "cache_write_input_tokens": 0,
104
+ "cached_input_tokens": 19968,
105
+ "input_tokens": 20264,
106
+ "output_tokens": 39,
107
+ "reasoning_output_tokens": 0
108
+ },
109
+ {
110
+ "cache_write_input_tokens": 0,
111
+ "cached_input_tokens": 20224,
112
+ "input_tokens": 20772,
113
+ "output_tokens": 42,
114
+ "reasoning_output_tokens": 0
115
+ },
116
+ {
117
+ "cache_write_input_tokens": 0,
118
+ "cached_input_tokens": 20608,
119
+ "input_tokens": 21024,
120
+ "output_tokens": 164,
121
+ "reasoning_output_tokens": 0
122
+ },
123
+ {
124
+ "cache_write_input_tokens": 0,
125
+ "cached_input_tokens": 20992,
126
+ "input_tokens": 21584,
127
+ "output_tokens": 39,
128
+ "reasoning_output_tokens": 0
129
+ },
130
+ {
131
+ "cache_write_input_tokens": 0,
132
+ "cached_input_tokens": 21504,
133
+ "input_tokens": 21855,
134
+ "output_tokens": 167,
135
+ "reasoning_output_tokens": 0
136
+ },
137
+ {
138
+ "cache_write_input_tokens": 0,
139
+ "cached_input_tokens": 21888,
140
+ "input_tokens": 23252,
141
+ "output_tokens": 406,
142
+ "reasoning_output_tokens": 26
143
+ },
144
+ {
145
+ "cache_write_input_tokens": 0,
146
+ "cached_input_tokens": 23552,
147
+ "input_tokens": 24103,
148
+ "output_tokens": 140,
149
+ "reasoning_output_tokens": 21
150
+ },
151
+ {
152
+ "cache_write_input_tokens": 0,
153
+ "cached_input_tokens": 24064,
154
+ "input_tokens": 26193,
155
+ "output_tokens": 361,
156
+ "reasoning_output_tokens": 25
157
+ },
158
+ {
159
+ "cache_write_input_tokens": 0,
160
+ "cached_input_tokens": 26368,
161
+ "input_tokens": 26857,
162
+ "output_tokens": 1241,
163
+ "reasoning_output_tokens": 258
164
+ },
165
+ {
166
+ "cache_write_input_tokens": 0,
167
+ "cached_input_tokens": 27904,
168
+ "input_tokens": 28121,
169
+ "output_tokens": 81,
170
+ "reasoning_output_tokens": 0
171
+ },
172
+ {
173
+ "cache_write_input_tokens": 0,
174
+ "cached_input_tokens": 28032,
175
+ "input_tokens": 28847,
176
+ "output_tokens": 87,
177
+ "reasoning_output_tokens": 0
178
+ },
179
+ {
180
+ "cache_write_input_tokens": 0,
181
+ "cached_input_tokens": 28800,
182
+ "input_tokens": 28961,
183
+ "output_tokens": 81,
184
+ "reasoning_output_tokens": 0
185
+ },
186
+ {
187
+ "cache_write_input_tokens": 0,
188
+ "cached_input_tokens": 28928,
189
+ "input_tokens": 29553,
190
+ "output_tokens": 98,
191
+ "reasoning_output_tokens": 14
192
+ },
193
+ {
194
+ "cache_write_input_tokens": 0,
195
+ "cached_input_tokens": 29440,
196
+ "input_tokens": 30159,
197
+ "output_tokens": 56,
198
+ "reasoning_output_tokens": 0
199
+ },
200
+ {
201
+ "cache_write_input_tokens": 0,
202
+ "cached_input_tokens": 30080,
203
+ "input_tokens": 30720,
204
+ "output_tokens": 254,
205
+ "reasoning_output_tokens": 229
206
+ },
207
+ {
208
+ "cache_write_input_tokens": 0,
209
+ "cached_input_tokens": 30848,
210
+ "input_tokens": 30992,
211
+ "output_tokens": 23,
212
+ "reasoning_output_tokens": 0
213
+ },
214
+ {
215
+ "cache_write_input_tokens": 0,
216
+ "cached_input_tokens": 30976,
217
+ "input_tokens": 31184,
218
+ "output_tokens": 288,
219
+ "reasoning_output_tokens": 174
220
+ },
221
+ {
222
+ "cache_write_input_tokens": 0,
223
+ "cached_input_tokens": 31360,
224
+ "input_tokens": 31984,
225
+ "output_tokens": 160,
226
+ "reasoning_output_tokens": 83
227
+ },
228
+ {
229
+ "cache_write_input_tokens": 0,
230
+ "cached_input_tokens": 32000,
231
+ "input_tokens": 32638,
232
+ "output_tokens": 56,
233
+ "reasoning_output_tokens": 21
234
+ },
235
+ {
236
+ "cache_write_input_tokens": 0,
237
+ "cached_input_tokens": 32512,
238
+ "input_tokens": 33487,
239
+ "output_tokens": 23,
240
+ "reasoning_output_tokens": 0
241
+ },
242
+ {
243
+ "cache_write_input_tokens": 0,
244
+ "cached_input_tokens": 33408,
245
+ "input_tokens": 33528,
246
+ "output_tokens": 23,
247
+ "reasoning_output_tokens": 0
248
+ },
249
+ {
250
+ "cache_write_input_tokens": 0,
251
+ "cached_input_tokens": 33408,
252
+ "input_tokens": 33721,
253
+ "output_tokens": 555,
254
+ "reasoning_output_tokens": 52
255
+ },
256
+ {
257
+ "cache_write_input_tokens": 0,
258
+ "cached_input_tokens": 34176,
259
+ "input_tokens": 34296,
260
+ "output_tokens": 40,
261
+ "reasoning_output_tokens": 0
262
+ },
263
+ {
264
+ "cache_write_input_tokens": 0,
265
+ "cached_input_tokens": 34176,
266
+ "input_tokens": 34670,
267
+ "output_tokens": 193,
268
+ "reasoning_output_tokens": 117
269
+ },
270
+ {
271
+ "cache_write_input_tokens": 0,
272
+ "cached_input_tokens": 34688,
273
+ "input_tokens": 35365,
274
+ "output_tokens": 47,
275
+ "reasoning_output_tokens": 13
276
+ },
277
+ {
278
+ "cache_write_input_tokens": 0,
279
+ "cached_input_tokens": 35200,
280
+ "input_tokens": 35474,
281
+ "output_tokens": 345,
282
+ "reasoning_output_tokens": 300
283
+ },
284
+ {
285
+ "cache_write_input_tokens": 0,
286
+ "cached_input_tokens": 35584,
287
+ "input_tokens": 36417,
288
+ "output_tokens": 453,
289
+ "reasoning_output_tokens": 373
290
+ },
291
+ {
292
+ "cache_write_input_tokens": 0,
293
+ "cached_input_tokens": 36736,
294
+ "input_tokens": 37369,
295
+ "output_tokens": 315,
296
+ "reasoning_output_tokens": 223
297
+ },
298
+ {
299
+ "cache_write_input_tokens": 0,
300
+ "cached_input_tokens": 37632,
301
+ "input_tokens": 38183,
302
+ "output_tokens": 59,
303
+ "reasoning_output_tokens": 28
304
+ },
305
+ {
306
+ "cache_write_input_tokens": 0,
307
+ "cached_input_tokens": 38144,
308
+ "input_tokens": 38304,
309
+ "output_tokens": 93,
310
+ "reasoning_output_tokens": 51
311
+ },
312
+ {
313
+ "cache_write_input_tokens": 0,
314
+ "cached_input_tokens": 38272,
315
+ "input_tokens": 38878,
316
+ "output_tokens": 201,
317
+ "reasoning_output_tokens": 80
318
+ },
319
+ {
320
+ "cache_write_input_tokens": 0,
321
+ "cached_input_tokens": 38912,
322
+ "input_tokens": 39581,
323
+ "output_tokens": 32,
324
+ "reasoning_output_tokens": 0
325
+ },
326
+ {
327
+ "cache_write_input_tokens": 0,
328
+ "cached_input_tokens": 39424,
329
+ "input_tokens": 40093,
330
+ "output_tokens": 128,
331
+ "reasoning_output_tokens": 44
332
+ },
333
+ {
334
+ "cache_write_input_tokens": 0,
335
+ "cached_input_tokens": 40064,
336
+ "input_tokens": 40714,
337
+ "output_tokens": 87,
338
+ "reasoning_output_tokens": 32
339
+ },
340
+ {
341
+ "cache_write_input_tokens": 0,
342
+ "cached_input_tokens": 40576,
343
+ "input_tokens": 41287,
344
+ "output_tokens": 105,
345
+ "reasoning_output_tokens": 0
346
+ },
347
+ {
348
+ "cache_write_input_tokens": 0,
349
+ "cached_input_tokens": 41216,
350
+ "input_tokens": 41902,
351
+ "output_tokens": 40,
352
+ "reasoning_output_tokens": 0
353
+ },
354
+ {
355
+ "cache_write_input_tokens": 0,
356
+ "cached_input_tokens": 41728,
357
+ "input_tokens": 42274,
358
+ "output_tokens": 111,
359
+ "reasoning_output_tokens": 36
360
+ },
361
+ {
362
+ "cache_write_input_tokens": 0,
363
+ "cached_input_tokens": 42240,
364
+ "input_tokens": 42883,
365
+ "output_tokens": 56,
366
+ "reasoning_output_tokens": 0
367
+ },
368
+ {
369
+ "cache_write_input_tokens": 0,
370
+ "cached_input_tokens": 42880,
371
+ "input_tokens": 43430,
372
+ "output_tokens": 54,
373
+ "reasoning_output_tokens": 0
374
+ },
375
+ {
376
+ "cache_write_input_tokens": 0,
377
+ "cached_input_tokens": 43392,
378
+ "input_tokens": 43967,
379
+ "output_tokens": 81,
380
+ "reasoning_output_tokens": 0
381
+ },
382
+ {
383
+ "cache_write_input_tokens": 0,
384
+ "cached_input_tokens": 43904,
385
+ "input_tokens": 44545,
386
+ "output_tokens": 66,
387
+ "reasoning_output_tokens": 32
388
+ },
389
+ {
390
+ "cache_write_input_tokens": 0,
391
+ "cached_input_tokens": 44416,
392
+ "input_tokens": 45087,
393
+ "output_tokens": 225,
394
+ "reasoning_output_tokens": 111
395
+ },
396
+ {
397
+ "cache_write_input_tokens": 0,
398
+ "cached_input_tokens": 45184,
399
+ "input_tokens": 45813,
400
+ "output_tokens": 101,
401
+ "reasoning_output_tokens": 29
402
+ },
403
+ {
404
+ "cache_write_input_tokens": 0,
405
+ "cached_input_tokens": 45696,
406
+ "input_tokens": 46415,
407
+ "output_tokens": 32,
408
+ "reasoning_output_tokens": 0
409
+ },
410
+ {
411
+ "cache_write_input_tokens": 0,
412
+ "cached_input_tokens": 46208,
413
+ "input_tokens": 46923,
414
+ "output_tokens": 112,
415
+ "reasoning_output_tokens": 54
416
+ },
417
+ {
418
+ "cache_write_input_tokens": 0,
419
+ "cached_input_tokens": 46976,
420
+ "input_tokens": 47169,
421
+ "output_tokens": 99,
422
+ "reasoning_output_tokens": 47
423
+ },
424
+ {
425
+ "cache_write_input_tokens": 0,
426
+ "cached_input_tokens": 47232,
427
+ "input_tokens": 47665,
428
+ "output_tokens": 745,
429
+ "reasoning_output_tokens": 154
430
+ },
431
+ {
432
+ "cache_write_input_tokens": 0,
433
+ "cached_input_tokens": 48256,
434
+ "input_tokens": 48433,
435
+ "output_tokens": 41,
436
+ "reasoning_output_tokens": 0
437
+ }
438
+ ],
439
+ "usage_source": "provider-raw"
440
+ },
441
+ "cost_note": "API equivalent sums the native Pi session cost estimates. Subscription billing is not reported.",
442
+ "media_note": "All three native views and every frame retained. Frame timing is unchanged; playback starts at zero.",
443
+ "artifacts": [
444
+ {
445
+ "name": "artifacts.json",
446
+ "path": "experiments/robotwin-native/kinex-1/artifacts.json",
447
+ "bytes": 2650,
448
+ "sha256": "e14a1c36466d235acb6a46678cba9b5a54b62e50e9af6e31fc55fb07e0a00902"
449
+ },
450
+ {
451
+ "name": "combined.mp4",
452
+ "path": "experiments/robotwin-native/kinex-1/combined.mp4",
453
+ "bytes": 4426774,
454
+ "sha256": "2c086a9ec449338ab3056eff3315773b24fa89b3737c81a31337ba11a74ed81f"
455
+ },
456
+ {
457
+ "name": "effective-model.json",
458
+ "path": "experiments/robotwin-native/kinex-1/effective-model.json",
459
+ "bytes": 444,
460
+ "sha256": "5a3694bfdd4fbeb4b1f47e125dde01703a7e004ec43f0bd6b47bbe7ba7c989e5"
461
+ },
462
+ {
463
+ "name": "episode-report.json",
464
+ "path": "experiments/robotwin-native/kinex-1/episode-report.json",
465
+ "bytes": 13825,
466
+ "sha256": "1238d688b661903e90bda73f73c6d6f502ade4f8a73d67b36719a5e6241b9347"
467
+ },
468
+ {
469
+ "name": "episode.json",
470
+ "path": "experiments/robotwin-native/kinex-1/episode.json",
471
+ "bytes": 279,
472
+ "sha256": "c227d427ec263488571a1e59c092b626a1077cc1f79ce5c82c0719ee833cab28"
473
+ },
474
+ {
475
+ "name": "evidence.tar.gz",
476
+ "path": "experiments/robotwin-native/kinex-1/evidence.tar.gz",
477
+ "bytes": 5690824,
478
+ "sha256": "66000d8bb81d51f2a4baecf42b7fcaaf237e23a1a544fe68e3c4e0f6073a9042"
479
+ },
480
+ {
481
+ "name": "instruction.json",
482
+ "path": "experiments/robotwin-native/kinex-1/instruction.json",
483
+ "bytes": 5227,
484
+ "sha256": "38ee681e1405cea1b9239be4f9f0578d200b344baa08fd6b7a83a03e05fdaac6"
485
+ },
486
+ {
487
+ "name": "media.json",
488
+ "path": "experiments/robotwin-native/kinex-1/media.json",
489
+ "bytes": 1796,
490
+ "sha256": "27c7301a5f97647ed62433d771786ba0949114a10b5800a5f983a471cf26a241"
491
+ },
492
+ {
493
+ "name": "observations/01d7a694-1.png",
494
+ "path": "experiments/robotwin-native/kinex-1/observations/01d7a694-1.png",
495
+ "bytes": 44485,
496
+ "sha256": "c2dc8518b35ec32c0843705f7d4f5bd002207d5d4a204241ac8a185c56fdc6f9"
497
+ },
498
+ {
499
+ "name": "observations/0bd66e5e-1.png",
500
+ "path": "experiments/robotwin-native/kinex-1/observations/0bd66e5e-1.png",
501
+ "bytes": 65531,
502
+ "sha256": "5898972bf8cd2f60b04f249d04a817cd22f220ebcd8df8dd83564973c4fe8ef1"
503
+ },
504
+ {
505
+ "name": "observations/0d7d46d8-1.png",
506
+ "path": "experiments/robotwin-native/kinex-1/observations/0d7d46d8-1.png",
507
+ "bytes": 73322,
508
+ "sha256": "967dbb5f94156cf332b4eb3b32b24de36315032c1f73acc2cdce937e509fcb38"
509
+ },
510
+ {
511
+ "name": "observations/13f789a0-1.png",
512
+ "path": "experiments/robotwin-native/kinex-1/observations/13f789a0-1.png",
513
+ "bytes": 57119,
514
+ "sha256": "2deb1358a0b58fdacc70b6a72505bafba5bab8fe7b8d8831bfaa2ca7665af2b8"
515
+ },
516
+ {
517
+ "name": "observations/1953a6e9-1.png",
518
+ "path": "experiments/robotwin-native/kinex-1/observations/1953a6e9-1.png",
519
+ "bytes": 57119,
520
+ "sha256": "2deb1358a0b58fdacc70b6a72505bafba5bab8fe7b8d8831bfaa2ca7665af2b8"
521
+ },
522
+ {
523
+ "name": "observations/245fef43-1.png",
524
+ "path": "experiments/robotwin-native/kinex-1/observations/245fef43-1.png",
525
+ "bytes": 73357,
526
+ "sha256": "b15c6eec1c027b907e4b5387689ac8819b1cc6b1f8420b703b49af1bea028e6a"
527
+ },
528
+ {
529
+ "name": "observations/3d9b5793-1.png",
530
+ "path": "experiments/robotwin-native/kinex-1/observations/3d9b5793-1.png",
531
+ "bytes": 72061,
532
+ "sha256": "8d33d3ca47c9abfe2504aaa4c76569dae7832430802cf1c2b3bc9d3a2473e38b"
533
+ },
534
+ {
535
+ "name": "observations/451da7eb-1.png",
536
+ "path": "experiments/robotwin-native/kinex-1/observations/451da7eb-1.png",
537
+ "bytes": 73322,
538
+ "sha256": "967dbb5f94156cf332b4eb3b32b24de36315032c1f73acc2cdce937e509fcb38"
539
+ },
540
+ {
541
+ "name": "observations/4a6cdfeb-1.png",
542
+ "path": "experiments/robotwin-native/kinex-1/observations/4a6cdfeb-1.png",
543
+ "bytes": 67527,
544
+ "sha256": "87b9f4f70f09c8a9425084e9a2d3e6a4ffc2520334ee63427910c31efa43706f"
545
+ },
546
+ {
547
+ "name": "observations/4f33bd50-1.png",
548
+ "path": "experiments/robotwin-native/kinex-1/observations/4f33bd50-1.png",
549
+ "bytes": 74189,
550
+ "sha256": "5973e4dabdfae8fcd915d6dcee6e96bf5c821ae5ddd7104678363f67208b9339"
551
+ },
552
+ {
553
+ "name": "observations/5c5beb92-1.png",
554
+ "path": "experiments/robotwin-native/kinex-1/observations/5c5beb92-1.png",
555
+ "bytes": 70312,
556
+ "sha256": "c80af286e96bb9d8396279e7d72e13e213878996968ddd1770befce3c7ba79fe"
557
+ },
558
+ {
559
+ "name": "observations/5df3e384-1.png",
560
+ "path": "experiments/robotwin-native/kinex-1/observations/5df3e384-1.png",
561
+ "bytes": 67527,
562
+ "sha256": "87b9f4f70f09c8a9425084e9a2d3e6a4ffc2520334ee63427910c31efa43706f"
563
+ },
564
+ {
565
+ "name": "observations/5ffc47a8-1.png",
566
+ "path": "experiments/robotwin-native/kinex-1/observations/5ffc47a8-1.png",
567
+ "bytes": 68076,
568
+ "sha256": "56378baf83446ea9e3ea504fc34bead3fe83a8d82363db3eee3f36b43c256c3e"
569
+ },
570
+ {
571
+ "name": "observations/797caa5e-1.png",
572
+ "path": "experiments/robotwin-native/kinex-1/observations/797caa5e-1.png",
573
+ "bytes": 69718,
574
+ "sha256": "9661a76f97939ef8faf0daf63fed74fc4f5096ef2e6d19ae0a0e2407f4a95912"
575
+ },
576
+ {
577
+ "name": "observations/86db988c-1.png",
578
+ "path": "experiments/robotwin-native/kinex-1/observations/86db988c-1.png",
579
+ "bytes": 49968,
580
+ "sha256": "ac8261e059c2fe5f5d7ac7edb3c71b5a1250a10693362fa5e74df6ab56e2230e"
581
+ },
582
+ {
583
+ "name": "observations/86db988c-2.png",
584
+ "path": "experiments/robotwin-native/kinex-1/observations/86db988c-2.png",
585
+ "bytes": 38472,
586
+ "sha256": "df6f4a7797be1a981859b4f8c9e592189d13d33735005ddf30031cdc79e88bd0"
587
+ },
588
+ {
589
+ "name": "observations/a51b5070-1.png",
590
+ "path": "experiments/robotwin-native/kinex-1/observations/a51b5070-1.png",
591
+ "bytes": 79008,
592
+ "sha256": "8e99ee84d9fd52305b8fe9592b87f3a5e4ef8ab559bd16f854d9c5e2fca832f2"
593
+ },
594
+ {
595
+ "name": "observations/a8adce0d-1.png",
596
+ "path": "experiments/robotwin-native/kinex-1/observations/a8adce0d-1.png",
597
+ "bytes": 65331,
598
+ "sha256": "88d22e07336e0c04517b637f0afa357436f1ce31c8e8477a5fae48a893bddd2b"
599
+ },
600
+ {
601
+ "name": "observations/af212c1a-1.png",
602
+ "path": "experiments/robotwin-native/kinex-1/observations/af212c1a-1.png",
603
+ "bytes": 71170,
604
+ "sha256": "1076a1d6b891d60e6e1091ec79ff81955098ea5a29f58f2cf6cb4f96770c0cf4"
605
+ },
606
+ {
607
+ "name": "observations/b5d4c986-1.png",
608
+ "path": "experiments/robotwin-native/kinex-1/observations/b5d4c986-1.png",
609
+ "bytes": 78083,
610
+ "sha256": "04e3f35744e4a017f4431374a4d6db2f8df752aec0049d69b2e065493d1db355"
611
+ },
612
+ {
613
+ "name": "observations/b833ee1b-1.png",
614
+ "path": "experiments/robotwin-native/kinex-1/observations/b833ee1b-1.png",
615
+ "bytes": 64020,
616
+ "sha256": "d8482ebbc4794fdcfd2c14e7a43a2478ae0c40ee0a77277fb6c8353654da843b"
617
+ },
618
+ {
619
+ "name": "observations/bab02154-1.png",
620
+ "path": "experiments/robotwin-native/kinex-1/observations/bab02154-1.png",
621
+ "bytes": 68601,
622
+ "sha256": "1cdc9df3bcc33fe6d3b95f2816df2089b490535803a81cbfe07e3329a8d0edc1"
623
+ },
624
+ {
625
+ "name": "observations/bf1286a6-1.png",
626
+ "path": "experiments/robotwin-native/kinex-1/observations/bf1286a6-1.png",
627
+ "bytes": 71170,
628
+ "sha256": "1076a1d6b891d60e6e1091ec79ff81955098ea5a29f58f2cf6cb4f96770c0cf4"
629
+ },
630
+ {
631
+ "name": "observations/c1d037d5-1.png",
632
+ "path": "experiments/robotwin-native/kinex-1/observations/c1d037d5-1.png",
633
+ "bytes": 40108,
634
+ "sha256": "a02773c8bf900439c7394853f9e9ed9e5190d12ad90596d50567b6aadac5e94d"
635
+ },
636
+ {
637
+ "name": "observations/c3226fa8-1.png",
638
+ "path": "experiments/robotwin-native/kinex-1/observations/c3226fa8-1.png",
639
+ "bytes": 78082,
640
+ "sha256": "788214a18267ec4b321f389b6e22fa52198b77e93f5af12a94fabcab40deda01"
641
+ },
642
+ {
643
+ "name": "observations/db1b4a52-1.png",
644
+ "path": "experiments/robotwin-native/kinex-1/observations/db1b4a52-1.png",
645
+ "bytes": 63997,
646
+ "sha256": "78812430933b28361b3a59b26481652cca938e3e0887582065f4c43ce0e2a11f"
647
+ },
648
+ {
649
+ "name": "observations/e7cd2d3f-1.png",
650
+ "path": "experiments/robotwin-native/kinex-1/observations/e7cd2d3f-1.png",
651
+ "bytes": 78989,
652
+ "sha256": "566ddf72eee2e6c80c626f25cb1ee1e8a4372080aa828815f27b12a62cbab687"
653
+ },
654
+ {
655
+ "name": "observations/e84d68b9-1.png",
656
+ "path": "experiments/robotwin-native/kinex-1/observations/e84d68b9-1.png",
657
+ "bytes": 72381,
658
+ "sha256": "04609cbef12ba796fb13673c6bcc28d3d92b5511739316561095562da7f6bfd7"
659
+ },
660
+ {
661
+ "name": "observations/f0765732-1.png",
662
+ "path": "experiments/robotwin-native/kinex-1/observations/f0765732-1.png",
663
+ "bytes": 78082,
664
+ "sha256": "788214a18267ec4b321f389b6e22fa52198b77e93f5af12a94fabcab40deda01"
665
+ },
666
+ {
667
+ "name": "observations/f9d2862d-1.png",
668
+ "path": "experiments/robotwin-native/kinex-1/observations/f9d2862d-1.png",
669
+ "bytes": 52369,
670
+ "sha256": "3112b4e27a82a72f6578443f826d8396ae43fad2f1ee267b1bff10fb3b802945"
671
+ },
672
+ {
673
+ "name": "policy-calls.json",
674
+ "path": "experiments/robotwin-native/kinex-1/policy-calls.json",
675
+ "bytes": 3,
676
+ "sha256": "37517e5f3dc66819f61f5a7bb8ace1921282415f10551d2defa5c3eb0985b570"
677
+ },
678
+ {
679
+ "name": "poster.jpg",
680
+ "path": "experiments/robotwin-native/kinex-1/poster.jpg",
681
+ "bytes": 21320,
682
+ "sha256": "4b102c75085f60b661af5234d659217e7ab89fd1106e0b531c9e8e0d54a59c4d"
683
+ },
684
+ {
685
+ "name": "provider-requests.jsonl",
686
+ "path": "experiments/robotwin-native/kinex-1/provider-requests.jsonl",
687
+ "bytes": 1047529,
688
+ "sha256": "76a2ae03215bda2c5165f8c10cec76b8635ca15ecfb2a6c10056fc8b35d2e2c8"
689
+ },
690
+ {
691
+ "name": "session.json",
692
+ "path": "experiments/robotwin-native/kinex-1/session.json",
693
+ "bytes": 123231,
694
+ "sha256": "3d33a8fca9c4bbe6659a4bc6b96643be7eb780587dd85edd8fbfb3b8ccce5239"
695
+ },
696
+ {
697
+ "name": "session.jsonl",
698
+ "path": "experiments/robotwin-native/kinex-1/session.jsonl",
699
+ "bytes": 2958277,
700
+ "sha256": "44f7beb27d0e5e6dae9af4c27f804d7bfffe16d03c93aa48102fd5d32fa20436"
701
+ },
702
+ {
703
+ "name": "terminal.jpg",
704
+ "path": "experiments/robotwin-native/kinex-1/terminal.jpg",
705
+ "bytes": 42796,
706
+ "sha256": "df7e3c67174889a8b50077c9993f463bfc820bba3765a3b20f83fbedf5b05efa"
707
+ },
708
+ {
709
+ "name": "tool-calls.json",
710
+ "path": "experiments/robotwin-native/kinex-1/tool-calls.json",
711
+ "bytes": 117023,
712
+ "sha256": "ce54431a54328524d86c6b2b6f7682db61da58f90d687d2a05c8375edea33314"
713
+ },
714
+ {
715
+ "name": "trace.json",
716
+ "path": "experiments/robotwin-native/kinex-1/trace.json",
717
+ "bytes": 234270,
718
+ "sha256": "141afab566147ae3f6e19acf2ccdf678ab12a92405c5a475fc6f55a50c334d53"
719
+ },
720
+ {
721
+ "name": "usage-audit.json",
722
+ "path": "experiments/robotwin-native/kinex-1/usage-audit.json",
723
+ "bytes": 11178,
724
+ "sha256": "0701261b643a2be34d2e7a178ca33f9442f8d66d6d0499545865fc4f0fda8874"
725
+ },
726
+ {
727
+ "name": "workspace/.kinex/devices/head_camera/profile.yaml",
728
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/devices/head_camera/profile.yaml",
729
+ "bytes": 2691,
730
+ "sha256": "743e9be788b817a2f90875746bc58b93de3f810eef6b720d93af522d6dcb664b"
731
+ },
732
+ {
733
+ "name": "workspace/.kinex/devices/left_arm/arm.urdf",
734
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/devices/left_arm/arm.urdf",
735
+ "bytes": 1879,
736
+ "sha256": "e53c41aaf34a08cb0224e5bf2bf9bbf949813018c2e0091b8021ef04d82a3f41"
737
+ },
738
+ {
739
+ "name": "workspace/.kinex/devices/left_arm/links.py",
740
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/devices/left_arm/links.py",
741
+ "bytes": 233,
742
+ "sha256": "6858db0b49065b895a79860ada359fd6da79a32e4c5b2950d159678420e85844"
743
+ },
744
+ {
745
+ "name": "workspace/.kinex/devices/left_arm/profile.yaml",
746
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/devices/left_arm/profile.yaml",
747
+ "bytes": 2714,
748
+ "sha256": "4083ed5aa12269768f74061b83bf18fdaa72c651351f12d9d20150372c74eb3d"
749
+ },
750
+ {
751
+ "name": "workspace/.kinex/devices/left_camera/profile.yaml",
752
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/devices/left_camera/profile.yaml",
753
+ "bytes": 2734,
754
+ "sha256": "3d4fa5b59734390f3019355b8b0260f796646a5b0639d0afe07bebdcb33475f9"
755
+ },
756
+ {
757
+ "name": "workspace/.kinex/devices/left_gripper/gripper.urdf",
758
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/devices/left_gripper/gripper.urdf",
759
+ "bytes": 1401,
760
+ "sha256": "7f5aad8ae559c5196bb28a9d72d7020bf5d7b79d5af49aefdb0e6ce279e1f4be"
761
+ },
762
+ {
763
+ "name": "workspace/.kinex/devices/left_gripper/links.py",
764
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/devices/left_gripper/links.py",
765
+ "bytes": 233,
766
+ "sha256": "6858db0b49065b895a79860ada359fd6da79a32e4c5b2950d159678420e85844"
767
+ },
768
+ {
769
+ "name": "workspace/.kinex/devices/left_gripper/profile.yaml",
770
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/devices/left_gripper/profile.yaml",
771
+ "bytes": 2594,
772
+ "sha256": "4a5eca3d3f3cf5f4dfa9e5b2527f02ee747354a371dc5dcd51731271685aa663"
773
+ },
774
+ {
775
+ "name": "workspace/.kinex/devices/right_arm/arm.urdf",
776
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/devices/right_arm/arm.urdf",
777
+ "bytes": 1881,
778
+ "sha256": "cb210121ba099341eeffdcc95588deaa0524037cff46602940d1037628600d8e"
779
+ },
780
+ {
781
+ "name": "workspace/.kinex/devices/right_arm/links.py",
782
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/devices/right_arm/links.py",
783
+ "bytes": 233,
784
+ "sha256": "6858db0b49065b895a79860ada359fd6da79a32e4c5b2950d159678420e85844"
785
+ },
786
+ {
787
+ "name": "workspace/.kinex/devices/right_arm/profile.yaml",
788
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/devices/right_arm/profile.yaml",
789
+ "bytes": 2728,
790
+ "sha256": "688fd66bd6f189ea56d74a66fb25bed3b4883667f5a46218145c22199d2038ca"
791
+ },
792
+ {
793
+ "name": "workspace/.kinex/devices/right_camera/profile.yaml",
794
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/devices/right_camera/profile.yaml",
795
+ "bytes": 2742,
796
+ "sha256": "439d6c17a9dd7b27310586ee88c31e111fb1183b4ecc11c1d94d3df508c23ea0"
797
+ },
798
+ {
799
+ "name": "workspace/.kinex/devices/right_gripper/gripper.urdf",
800
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/devices/right_gripper/gripper.urdf",
801
+ "bytes": 1407,
802
+ "sha256": "347f6969d8189abeee10b5a3e5358b91511858da9fb233cdf062f80c53b97dd9"
803
+ },
804
+ {
805
+ "name": "workspace/.kinex/devices/right_gripper/links.py",
806
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/devices/right_gripper/links.py",
807
+ "bytes": 233,
808
+ "sha256": "6858db0b49065b895a79860ada359fd6da79a32e4c5b2950d159678420e85844"
809
+ },
810
+ {
811
+ "name": "workspace/.kinex/devices/right_gripper/profile.yaml",
812
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/devices/right_gripper/profile.yaml",
813
+ "bytes": 2614,
814
+ "sha256": "b82ca28d355d37c54e71c32aa3ddcf9097caf7a49d00b7b4ce0f26d96d6019b2"
815
+ },
816
+ {
817
+ "name": "workspace/.kinex/devices/world/profile.yaml",
818
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/devices/world/profile.yaml",
819
+ "bytes": 1593,
820
+ "sha256": "892dc5a6883ee19b1c5624bcdd6a30ae427f81b5b660292eaacd18eb79497ebd"
821
+ },
822
+ {
823
+ "name": "workspace/.kinex/kinex.json",
824
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/kinex.json",
825
+ "bytes": 50,
826
+ "sha256": "703ec9cee9365c7c5c4b967219de9a9e7b9ca17517eae82cf7af30431de4026e"
827
+ },
828
+ {
829
+ "name": "workspace/.kinex/memos/handover-block.md",
830
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/memos/handover-block.md",
831
+ "bytes": 2335,
832
+ "sha256": "0ecd8b2b4b1c48092e2bb5276168f46558522b184727c5f355aad9ba9fd47397"
833
+ },
834
+ {
835
+ "name": "workspace/.kinex/tools/cartesian_move.py",
836
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/tools/cartesian_move.py",
837
+ "bytes": 3118,
838
+ "sha256": "3f9739a26a7407857e75f40465c9f6e46bf6aba7f177f540a99e7e14af27b184"
839
+ },
840
+ {
841
+ "name": "workspace/.kinex/tools/color_geometry.py",
842
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/tools/color_geometry.py",
843
+ "bytes": 1457,
844
+ "sha256": "bee598b8704d3408de37ed15f331d291b79959ff7dce6684aa2cca3ec43c8a25"
845
+ },
846
+ {
847
+ "name": "workspace/.kinex/workspace.yaml",
848
+ "path": "experiments/robotwin-native/kinex-1/workspace/.kinex/workspace.yaml",
849
+ "bytes": 335,
850
+ "sha256": "8a0cbe7e2438339e12693ecdda4530b8601e092a11c9e1d5f6301bd67a415ca7"
851
+ },
852
+ {
853
+ "name": "workspace/harness.md",
854
+ "path": "experiments/robotwin-native/kinex-1/workspace/harness.md",
855
+ "bytes": 444,
856
+ "sha256": "81c1ef513247775681eb49b5ac07a5f4eebc7848afab963ef21a3bca473b8485"
857
+ },
858
+ {
859
+ "name": "workspace/instruction.md",
860
+ "path": "experiments/robotwin-native/kinex-1/workspace/instruction.md",
861
+ "bytes": 2238,
862
+ "sha256": "58a832df9bc84bf4ba537a1c3e6afe9b35e54c78a52d9417e7bb11b6d4177734"
863
+ },
864
+ {
865
+ "name": "workspace.json",
866
+ "path": "experiments/robotwin-native/kinex-1/workspace.json",
867
+ "bytes": 4817,
868
+ "sha256": "fb8fcc2563b6b61e0ab2cf0b8426297949ddc9e397a19d0f007799dba5a7ff33"
869
+ },
870
+ {
871
+ "name": "workspace.tar.gz",
872
+ "path": "experiments/robotwin-native/kinex-1/workspace.tar.gz",
873
+ "bytes": 9698,
874
+ "sha256": "fd41396cf2740d7a4ce508ee2b18adbec370d293c8f80c1e766b7520047ef36f"
875
+ }
876
+ ],
877
+ "native": {
878
+ "episode": 1,
879
+ "initialization": {
880
+ "expert_plan_success": false,
881
+ "requested_seed": 0
882
+ },
883
+ "native_reward": 1.0,
884
+ "planned_episodes": 1,
885
+ "reason": "success",
886
+ "simulation_time_s": 24.84,
887
+ "step_limit": 800,
888
+ "steps": 621,
889
+ "success": true,
890
+ "valid": true,
891
+ "step": "episode_01",
892
+ "evidence_valid": true,
893
+ "wall_time_sec": 635.897416,
894
+ "usage": {
895
+ "n_input_tokens": 1774043,
896
+ "n_output_tokens": 8679,
897
+ "n_cache_tokens": 1715200,
898
+ "cost_usd": null,
899
+ "n_fresh_input_tokens": 58843,
900
+ "n_cache_read_tokens": 1715200,
901
+ "n_cache_write_tokens": 0
902
+ },
903
+ "usage_details": {
904
+ "accounting": "reported-responses",
905
+ "audit_complete": true,
906
+ "audit_source": "Provider terminal response usage",
907
+ "cache_hit_rate": 0.9668311309252369,
908
+ "cache_read_input_tokens": 1715200,
909
+ "cache_reported_input_tokens": 1774043,
910
+ "cache_write_basis": "provider-reported",
911
+ "cache_write_input_tokens": 0,
912
+ "cache_write_reported_input_tokens": 1774043,
913
+ "cached_input_tokens": 1715200,
914
+ "cost_usd": null,
915
+ "fresh_input_tokens": 58843,
916
+ "harbor_mismatches": {},
917
+ "input_breakdown_complete": true,
918
+ "input_tokens": 1774043,
919
+ "known_cache_write_input_tokens": 0,
920
+ "known_cached_input_tokens": 1715200,
921
+ "known_fresh_input_tokens": 58843,
922
+ "known_input_tokens": 1774043,
923
+ "known_output_tokens": 8679,
924
+ "known_reasoning_output_tokens": 2627,
925
+ "output_tokens": 8679,
926
+ "parser": "harbor/0.22.0",
927
+ "reasoning_output_tokens": 2627,
928
+ "reasoning_reported_output_tokens": 8679,
929
+ "rejected_attempts": 0,
930
+ "reported_responses": {
931
+ "cache_reported_input_tokens": 54,
932
+ "cache_write_input_tokens": 54,
933
+ "cache_write_reported_input_tokens": 54,
934
+ "cached_input_tokens": 54,
935
+ "fresh_input_tokens": 54,
936
+ "input_tokens": 54,
937
+ "output_tokens": 54,
938
+ "reasoning_output_tokens": 54,
939
+ "reasoning_reported_output_tokens": 54
940
+ },
941
+ "request_attempts": 54,
942
+ "response_count": 54,
943
+ "response_ids_complete": true,
944
+ "schema": "rlebench/token-usage/1",
945
+ "source": "Harbor session metrics",
946
+ "total_tokens": 1782722,
947
+ "uncached_input_tokens": 58843,
948
+ "unresolved_attempts": [],
949
+ "usage_records": [
950
+ {
951
+ "cache_write_input_tokens": 0,
952
+ "cached_input_tokens": 0,
953
+ "input_tokens": 11322,
954
+ "output_tokens": 19,
955
+ "reasoning_output_tokens": 0
956
+ },
957
+ {
958
+ "cache_write_input_tokens": 0,
959
+ "cached_input_tokens": 11136,
960
+ "input_tokens": 11362,
961
+ "output_tokens": 29,
962
+ "reasoning_output_tokens": 0
963
+ },
964
+ {
965
+ "cache_write_input_tokens": 0,
966
+ "cached_input_tokens": 0,
967
+ "input_tokens": 11495,
968
+ "output_tokens": 30,
969
+ "reasoning_output_tokens": 0
970
+ },
971
+ {
972
+ "cache_write_input_tokens": 0,
973
+ "cached_input_tokens": 11392,
974
+ "input_tokens": 14101,
975
+ "output_tokens": 223,
976
+ "reasoning_output_tokens": 0
977
+ },
978
+ {
979
+ "cache_write_input_tokens": 0,
980
+ "cached_input_tokens": 14208,
981
+ "input_tokens": 18793,
982
+ "output_tokens": 55,
983
+ "reasoning_output_tokens": 0
984
+ },
985
+ {
986
+ "cache_write_input_tokens": 0,
987
+ "cached_input_tokens": 18688,
988
+ "input_tokens": 20034,
989
+ "output_tokens": 78,
990
+ "reasoning_output_tokens": 0
991
+ },
992
+ {
993
+ "cache_write_input_tokens": 0,
994
+ "cached_input_tokens": 19968,
995
+ "input_tokens": 20264,
996
+ "output_tokens": 39,
997
+ "reasoning_output_tokens": 0
998
+ },
999
+ {
1000
+ "cache_write_input_tokens": 0,
1001
+ "cached_input_tokens": 20224,
1002
+ "input_tokens": 20772,
1003
+ "output_tokens": 42,
1004
+ "reasoning_output_tokens": 0
1005
+ },
1006
+ {
1007
+ "cache_write_input_tokens": 0,
1008
+ "cached_input_tokens": 20608,
1009
+ "input_tokens": 21024,
1010
+ "output_tokens": 164,
1011
+ "reasoning_output_tokens": 0
1012
+ },
1013
+ {
1014
+ "cache_write_input_tokens": 0,
1015
+ "cached_input_tokens": 20992,
1016
+ "input_tokens": 21584,
1017
+ "output_tokens": 39,
1018
+ "reasoning_output_tokens": 0
1019
+ },
1020
+ {
1021
+ "cache_write_input_tokens": 0,
1022
+ "cached_input_tokens": 21504,
1023
+ "input_tokens": 21855,
1024
+ "output_tokens": 167,
1025
+ "reasoning_output_tokens": 0
1026
+ },
1027
+ {
1028
+ "cache_write_input_tokens": 0,
1029
+ "cached_input_tokens": 21888,
1030
+ "input_tokens": 23252,
1031
+ "output_tokens": 406,
1032
+ "reasoning_output_tokens": 26
1033
+ },
1034
+ {
1035
+ "cache_write_input_tokens": 0,
1036
+ "cached_input_tokens": 23552,
1037
+ "input_tokens": 24103,
1038
+ "output_tokens": 140,
1039
+ "reasoning_output_tokens": 21
1040
+ },
1041
+ {
1042
+ "cache_write_input_tokens": 0,
1043
+ "cached_input_tokens": 24064,
1044
+ "input_tokens": 26193,
1045
+ "output_tokens": 361,
1046
+ "reasoning_output_tokens": 25
1047
+ },
1048
+ {
1049
+ "cache_write_input_tokens": 0,
1050
+ "cached_input_tokens": 26368,
1051
+ "input_tokens": 26857,
1052
+ "output_tokens": 1241,
1053
+ "reasoning_output_tokens": 258
1054
+ },
1055
+ {
1056
+ "cache_write_input_tokens": 0,
1057
+ "cached_input_tokens": 27904,
1058
+ "input_tokens": 28121,
1059
+ "output_tokens": 81,
1060
+ "reasoning_output_tokens": 0
1061
+ },
1062
+ {
1063
+ "cache_write_input_tokens": 0,
1064
+ "cached_input_tokens": 28032,
1065
+ "input_tokens": 28847,
1066
+ "output_tokens": 87,
1067
+ "reasoning_output_tokens": 0
1068
+ },
1069
+ {
1070
+ "cache_write_input_tokens": 0,
1071
+ "cached_input_tokens": 28800,
1072
+ "input_tokens": 28961,
1073
+ "output_tokens": 81,
1074
+ "reasoning_output_tokens": 0
1075
+ },
1076
+ {
1077
+ "cache_write_input_tokens": 0,
1078
+ "cached_input_tokens": 28928,
1079
+ "input_tokens": 29553,
1080
+ "output_tokens": 98,
1081
+ "reasoning_output_tokens": 14
1082
+ },
1083
+ {
1084
+ "cache_write_input_tokens": 0,
1085
+ "cached_input_tokens": 29440,
1086
+ "input_tokens": 30159,
1087
+ "output_tokens": 56,
1088
+ "reasoning_output_tokens": 0
1089
+ },
1090
+ {
1091
+ "cache_write_input_tokens": 0,
1092
+ "cached_input_tokens": 30080,
1093
+ "input_tokens": 30720,
1094
+ "output_tokens": 254,
1095
+ "reasoning_output_tokens": 229
1096
+ },
1097
+ {
1098
+ "cache_write_input_tokens": 0,
1099
+ "cached_input_tokens": 30848,
1100
+ "input_tokens": 30992,
1101
+ "output_tokens": 23,
1102
+ "reasoning_output_tokens": 0
1103
+ },
1104
+ {
1105
+ "cache_write_input_tokens": 0,
1106
+ "cached_input_tokens": 30976,
1107
+ "input_tokens": 31184,
1108
+ "output_tokens": 288,
1109
+ "reasoning_output_tokens": 174
1110
+ },
1111
+ {
1112
+ "cache_write_input_tokens": 0,
1113
+ "cached_input_tokens": 31360,
1114
+ "input_tokens": 31984,
1115
+ "output_tokens": 160,
1116
+ "reasoning_output_tokens": 83
1117
+ },
1118
+ {
1119
+ "cache_write_input_tokens": 0,
1120
+ "cached_input_tokens": 32000,
1121
+ "input_tokens": 32638,
1122
+ "output_tokens": 56,
1123
+ "reasoning_output_tokens": 21
1124
+ },
1125
+ {
1126
+ "cache_write_input_tokens": 0,
1127
+ "cached_input_tokens": 32512,
1128
+ "input_tokens": 33487,
1129
+ "output_tokens": 23,
1130
+ "reasoning_output_tokens": 0
1131
+ },
1132
+ {
1133
+ "cache_write_input_tokens": 0,
1134
+ "cached_input_tokens": 33408,
1135
+ "input_tokens": 33528,
1136
+ "output_tokens": 23,
1137
+ "reasoning_output_tokens": 0
1138
+ },
1139
+ {
1140
+ "cache_write_input_tokens": 0,
1141
+ "cached_input_tokens": 33408,
1142
+ "input_tokens": 33721,
1143
+ "output_tokens": 555,
1144
+ "reasoning_output_tokens": 52
1145
+ },
1146
+ {
1147
+ "cache_write_input_tokens": 0,
1148
+ "cached_input_tokens": 34176,
1149
+ "input_tokens": 34296,
1150
+ "output_tokens": 40,
1151
+ "reasoning_output_tokens": 0
1152
+ },
1153
+ {
1154
+ "cache_write_input_tokens": 0,
1155
+ "cached_input_tokens": 34176,
1156
+ "input_tokens": 34670,
1157
+ "output_tokens": 193,
1158
+ "reasoning_output_tokens": 117
1159
+ },
1160
+ {
1161
+ "cache_write_input_tokens": 0,
1162
+ "cached_input_tokens": 34688,
1163
+ "input_tokens": 35365,
1164
+ "output_tokens": 47,
1165
+ "reasoning_output_tokens": 13
1166
+ },
1167
+ {
1168
+ "cache_write_input_tokens": 0,
1169
+ "cached_input_tokens": 35200,
1170
+ "input_tokens": 35474,
1171
+ "output_tokens": 345,
1172
+ "reasoning_output_tokens": 300
1173
+ },
1174
+ {
1175
+ "cache_write_input_tokens": 0,
1176
+ "cached_input_tokens": 35584,
1177
+ "input_tokens": 36417,
1178
+ "output_tokens": 453,
1179
+ "reasoning_output_tokens": 373
1180
+ },
1181
+ {
1182
+ "cache_write_input_tokens": 0,
1183
+ "cached_input_tokens": 36736,
1184
+ "input_tokens": 37369,
1185
+ "output_tokens": 315,
1186
+ "reasoning_output_tokens": 223
1187
+ },
1188
+ {
1189
+ "cache_write_input_tokens": 0,
1190
+ "cached_input_tokens": 37632,
1191
+ "input_tokens": 38183,
1192
+ "output_tokens": 59,
1193
+ "reasoning_output_tokens": 28
1194
+ },
1195
+ {
1196
+ "cache_write_input_tokens": 0,
1197
+ "cached_input_tokens": 38144,
1198
+ "input_tokens": 38304,
1199
+ "output_tokens": 93,
1200
+ "reasoning_output_tokens": 51
1201
+ },
1202
+ {
1203
+ "cache_write_input_tokens": 0,
1204
+ "cached_input_tokens": 38272,
1205
+ "input_tokens": 38878,
1206
+ "output_tokens": 201,
1207
+ "reasoning_output_tokens": 80
1208
+ },
1209
+ {
1210
+ "cache_write_input_tokens": 0,
1211
+ "cached_input_tokens": 38912,
1212
+ "input_tokens": 39581,
1213
+ "output_tokens": 32,
1214
+ "reasoning_output_tokens": 0
1215
+ },
1216
+ {
1217
+ "cache_write_input_tokens": 0,
1218
+ "cached_input_tokens": 39424,
1219
+ "input_tokens": 40093,
1220
+ "output_tokens": 128,
1221
+ "reasoning_output_tokens": 44
1222
+ },
1223
+ {
1224
+ "cache_write_input_tokens": 0,
1225
+ "cached_input_tokens": 40064,
1226
+ "input_tokens": 40714,
1227
+ "output_tokens": 87,
1228
+ "reasoning_output_tokens": 32
1229
+ },
1230
+ {
1231
+ "cache_write_input_tokens": 0,
1232
+ "cached_input_tokens": 40576,
1233
+ "input_tokens": 41287,
1234
+ "output_tokens": 105,
1235
+ "reasoning_output_tokens": 0
1236
+ },
1237
+ {
1238
+ "cache_write_input_tokens": 0,
1239
+ "cached_input_tokens": 41216,
1240
+ "input_tokens": 41902,
1241
+ "output_tokens": 40,
1242
+ "reasoning_output_tokens": 0
1243
+ },
1244
+ {
1245
+ "cache_write_input_tokens": 0,
1246
+ "cached_input_tokens": 41728,
1247
+ "input_tokens": 42274,
1248
+ "output_tokens": 111,
1249
+ "reasoning_output_tokens": 36
1250
+ },
1251
+ {
1252
+ "cache_write_input_tokens": 0,
1253
+ "cached_input_tokens": 42240,
1254
+ "input_tokens": 42883,
1255
+ "output_tokens": 56,
1256
+ "reasoning_output_tokens": 0
1257
+ },
1258
+ {
1259
+ "cache_write_input_tokens": 0,
1260
+ "cached_input_tokens": 42880,
1261
+ "input_tokens": 43430,
1262
+ "output_tokens": 54,
1263
+ "reasoning_output_tokens": 0
1264
+ },
1265
+ {
1266
+ "cache_write_input_tokens": 0,
1267
+ "cached_input_tokens": 43392,
1268
+ "input_tokens": 43967,
1269
+ "output_tokens": 81,
1270
+ "reasoning_output_tokens": 0
1271
+ },
1272
+ {
1273
+ "cache_write_input_tokens": 0,
1274
+ "cached_input_tokens": 43904,
1275
+ "input_tokens": 44545,
1276
+ "output_tokens": 66,
1277
+ "reasoning_output_tokens": 32
1278
+ },
1279
+ {
1280
+ "cache_write_input_tokens": 0,
1281
+ "cached_input_tokens": 44416,
1282
+ "input_tokens": 45087,
1283
+ "output_tokens": 225,
1284
+ "reasoning_output_tokens": 111
1285
+ },
1286
+ {
1287
+ "cache_write_input_tokens": 0,
1288
+ "cached_input_tokens": 45184,
1289
+ "input_tokens": 45813,
1290
+ "output_tokens": 101,
1291
+ "reasoning_output_tokens": 29
1292
+ },
1293
+ {
1294
+ "cache_write_input_tokens": 0,
1295
+ "cached_input_tokens": 45696,
1296
+ "input_tokens": 46415,
1297
+ "output_tokens": 32,
1298
+ "reasoning_output_tokens": 0
1299
+ },
1300
+ {
1301
+ "cache_write_input_tokens": 0,
1302
+ "cached_input_tokens": 46208,
1303
+ "input_tokens": 46923,
1304
+ "output_tokens": 112,
1305
+ "reasoning_output_tokens": 54
1306
+ },
1307
+ {
1308
+ "cache_write_input_tokens": 0,
1309
+ "cached_input_tokens": 46976,
1310
+ "input_tokens": 47169,
1311
+ "output_tokens": 99,
1312
+ "reasoning_output_tokens": 47
1313
+ },
1314
+ {
1315
+ "cache_write_input_tokens": 0,
1316
+ "cached_input_tokens": 47232,
1317
+ "input_tokens": 47665,
1318
+ "output_tokens": 745,
1319
+ "reasoning_output_tokens": 154
1320
+ },
1321
+ {
1322
+ "cache_write_input_tokens": 0,
1323
+ "cached_input_tokens": 48256,
1324
+ "input_tokens": 48433,
1325
+ "output_tokens": 41,
1326
+ "reasoning_output_tokens": 0
1327
+ }
1328
+ ],
1329
+ "usage_source": "provider-raw"
1330
+ },
1331
+ "call_activity": {
1332
+ "schema": "rlebench/call-activity/1",
1333
+ "source": "Kinex native trace",
1334
+ "model_tool_calls": 61,
1335
+ "model_tool_calls_by_name": {
1336
+ "ls": 1,
1337
+ "read": 15,
1338
+ "find": 1,
1339
+ "exec": 32,
1340
+ "devices": 2,
1341
+ "write": 3,
1342
+ "edit": 1,
1343
+ "device": 4,
1344
+ "device_stop": 2
1345
+ },
1346
+ "python_instrumented_model_tool_calls": null,
1347
+ "python_tool_invocations": null,
1348
+ "nested_python_tool_invocations": null,
1349
+ "python_tool_invocations_by_origin": null,
1350
+ "python_device_rpc_attempts": null,
1351
+ "python_device_rpc_attempts_by_action": null,
1352
+ "python_device_rpc_errors": null,
1353
+ "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
1354
+ },
1355
+ "model_response_count": 54,
1356
+ "usage_source": "Harbor session metrics",
1357
+ "execution": {
1358
+ "status": "finished",
1359
+ "reason": null
1360
+ },
1361
+ "native": {
1362
+ "evidence_valid": true,
1363
+ "success": true,
1364
+ "termination": "success",
1365
+ "steps": 621
1366
+ },
1367
+ "failure_attribution": {
1368
+ "classification": "native_success",
1369
+ "physical_cause": null,
1370
+ "physical_cause_evidence": null
1371
+ },
1372
+ "exception": null
1373
+ },
1374
+ "model_call_count": 0
1375
+ }
experiments/robotwin-native/kinex-1/effective-model.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "kinex_identity": "74635879f639d2bf8918b7fab95bf1b37e29c187d24b1ea5692f22079d81d24f",
3
+ "kinex_version": "1.1.0",
4
+ "kinex_source": null,
5
+ "model": "openai-codex/gpt-6-astra",
6
+ "effort": "high",
7
+ "service_tier": null,
8
+ "fresh_session": true,
9
+ "max_steps": null,
10
+ "model_recording": false,
11
+ "compaction": "native",
12
+ "transport": "sse",
13
+ "strict_usage": true,
14
+ "max_request_retries": 50,
15
+ "usage_accounting": "reported-responses"
16
+ }
experiments/robotwin-native/kinex-1/episode-report.json ADDED
@@ -0,0 +1,497 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "episode": 1,
3
+ "initialization": {
4
+ "expert_plan_success": false,
5
+ "requested_seed": 0
6
+ },
7
+ "native_reward": 1.0,
8
+ "planned_episodes": 1,
9
+ "reason": "success",
10
+ "simulation_time_s": 24.84,
11
+ "step_limit": 800,
12
+ "steps": 621,
13
+ "success": true,
14
+ "valid": true,
15
+ "step": "episode_01",
16
+ "evidence_valid": true,
17
+ "wall_time_sec": 635.897416,
18
+ "usage": {
19
+ "n_input_tokens": 1774043,
20
+ "n_output_tokens": 8679,
21
+ "n_cache_tokens": 1715200,
22
+ "cost_usd": null,
23
+ "n_fresh_input_tokens": 58843,
24
+ "n_cache_read_tokens": 1715200,
25
+ "n_cache_write_tokens": 0
26
+ },
27
+ "usage_details": {
28
+ "accounting": "reported-responses",
29
+ "audit_complete": true,
30
+ "audit_source": "Provider terminal response usage",
31
+ "cache_hit_rate": 0.9668311309252369,
32
+ "cache_read_input_tokens": 1715200,
33
+ "cache_reported_input_tokens": 1774043,
34
+ "cache_write_basis": "provider-reported",
35
+ "cache_write_input_tokens": 0,
36
+ "cache_write_reported_input_tokens": 1774043,
37
+ "cached_input_tokens": 1715200,
38
+ "cost_usd": null,
39
+ "fresh_input_tokens": 58843,
40
+ "harbor_mismatches": {},
41
+ "input_breakdown_complete": true,
42
+ "input_tokens": 1774043,
43
+ "known_cache_write_input_tokens": 0,
44
+ "known_cached_input_tokens": 1715200,
45
+ "known_fresh_input_tokens": 58843,
46
+ "known_input_tokens": 1774043,
47
+ "known_output_tokens": 8679,
48
+ "known_reasoning_output_tokens": 2627,
49
+ "output_tokens": 8679,
50
+ "parser": "harbor/0.22.0",
51
+ "reasoning_output_tokens": 2627,
52
+ "reasoning_reported_output_tokens": 8679,
53
+ "rejected_attempts": 0,
54
+ "reported_responses": {
55
+ "cache_reported_input_tokens": 54,
56
+ "cache_write_input_tokens": 54,
57
+ "cache_write_reported_input_tokens": 54,
58
+ "cached_input_tokens": 54,
59
+ "fresh_input_tokens": 54,
60
+ "input_tokens": 54,
61
+ "output_tokens": 54,
62
+ "reasoning_output_tokens": 54,
63
+ "reasoning_reported_output_tokens": 54
64
+ },
65
+ "request_attempts": 54,
66
+ "response_count": 54,
67
+ "response_ids_complete": true,
68
+ "schema": "rlebench/token-usage/1",
69
+ "source": "Harbor session metrics",
70
+ "total_tokens": 1782722,
71
+ "uncached_input_tokens": 58843,
72
+ "unresolved_attempts": [],
73
+ "usage_records": [
74
+ {
75
+ "cache_write_input_tokens": 0,
76
+ "cached_input_tokens": 0,
77
+ "input_tokens": 11322,
78
+ "output_tokens": 19,
79
+ "reasoning_output_tokens": 0
80
+ },
81
+ {
82
+ "cache_write_input_tokens": 0,
83
+ "cached_input_tokens": 11136,
84
+ "input_tokens": 11362,
85
+ "output_tokens": 29,
86
+ "reasoning_output_tokens": 0
87
+ },
88
+ {
89
+ "cache_write_input_tokens": 0,
90
+ "cached_input_tokens": 0,
91
+ "input_tokens": 11495,
92
+ "output_tokens": 30,
93
+ "reasoning_output_tokens": 0
94
+ },
95
+ {
96
+ "cache_write_input_tokens": 0,
97
+ "cached_input_tokens": 11392,
98
+ "input_tokens": 14101,
99
+ "output_tokens": 223,
100
+ "reasoning_output_tokens": 0
101
+ },
102
+ {
103
+ "cache_write_input_tokens": 0,
104
+ "cached_input_tokens": 14208,
105
+ "input_tokens": 18793,
106
+ "output_tokens": 55,
107
+ "reasoning_output_tokens": 0
108
+ },
109
+ {
110
+ "cache_write_input_tokens": 0,
111
+ "cached_input_tokens": 18688,
112
+ "input_tokens": 20034,
113
+ "output_tokens": 78,
114
+ "reasoning_output_tokens": 0
115
+ },
116
+ {
117
+ "cache_write_input_tokens": 0,
118
+ "cached_input_tokens": 19968,
119
+ "input_tokens": 20264,
120
+ "output_tokens": 39,
121
+ "reasoning_output_tokens": 0
122
+ },
123
+ {
124
+ "cache_write_input_tokens": 0,
125
+ "cached_input_tokens": 20224,
126
+ "input_tokens": 20772,
127
+ "output_tokens": 42,
128
+ "reasoning_output_tokens": 0
129
+ },
130
+ {
131
+ "cache_write_input_tokens": 0,
132
+ "cached_input_tokens": 20608,
133
+ "input_tokens": 21024,
134
+ "output_tokens": 164,
135
+ "reasoning_output_tokens": 0
136
+ },
137
+ {
138
+ "cache_write_input_tokens": 0,
139
+ "cached_input_tokens": 20992,
140
+ "input_tokens": 21584,
141
+ "output_tokens": 39,
142
+ "reasoning_output_tokens": 0
143
+ },
144
+ {
145
+ "cache_write_input_tokens": 0,
146
+ "cached_input_tokens": 21504,
147
+ "input_tokens": 21855,
148
+ "output_tokens": 167,
149
+ "reasoning_output_tokens": 0
150
+ },
151
+ {
152
+ "cache_write_input_tokens": 0,
153
+ "cached_input_tokens": 21888,
154
+ "input_tokens": 23252,
155
+ "output_tokens": 406,
156
+ "reasoning_output_tokens": 26
157
+ },
158
+ {
159
+ "cache_write_input_tokens": 0,
160
+ "cached_input_tokens": 23552,
161
+ "input_tokens": 24103,
162
+ "output_tokens": 140,
163
+ "reasoning_output_tokens": 21
164
+ },
165
+ {
166
+ "cache_write_input_tokens": 0,
167
+ "cached_input_tokens": 24064,
168
+ "input_tokens": 26193,
169
+ "output_tokens": 361,
170
+ "reasoning_output_tokens": 25
171
+ },
172
+ {
173
+ "cache_write_input_tokens": 0,
174
+ "cached_input_tokens": 26368,
175
+ "input_tokens": 26857,
176
+ "output_tokens": 1241,
177
+ "reasoning_output_tokens": 258
178
+ },
179
+ {
180
+ "cache_write_input_tokens": 0,
181
+ "cached_input_tokens": 27904,
182
+ "input_tokens": 28121,
183
+ "output_tokens": 81,
184
+ "reasoning_output_tokens": 0
185
+ },
186
+ {
187
+ "cache_write_input_tokens": 0,
188
+ "cached_input_tokens": 28032,
189
+ "input_tokens": 28847,
190
+ "output_tokens": 87,
191
+ "reasoning_output_tokens": 0
192
+ },
193
+ {
194
+ "cache_write_input_tokens": 0,
195
+ "cached_input_tokens": 28800,
196
+ "input_tokens": 28961,
197
+ "output_tokens": 81,
198
+ "reasoning_output_tokens": 0
199
+ },
200
+ {
201
+ "cache_write_input_tokens": 0,
202
+ "cached_input_tokens": 28928,
203
+ "input_tokens": 29553,
204
+ "output_tokens": 98,
205
+ "reasoning_output_tokens": 14
206
+ },
207
+ {
208
+ "cache_write_input_tokens": 0,
209
+ "cached_input_tokens": 29440,
210
+ "input_tokens": 30159,
211
+ "output_tokens": 56,
212
+ "reasoning_output_tokens": 0
213
+ },
214
+ {
215
+ "cache_write_input_tokens": 0,
216
+ "cached_input_tokens": 30080,
217
+ "input_tokens": 30720,
218
+ "output_tokens": 254,
219
+ "reasoning_output_tokens": 229
220
+ },
221
+ {
222
+ "cache_write_input_tokens": 0,
223
+ "cached_input_tokens": 30848,
224
+ "input_tokens": 30992,
225
+ "output_tokens": 23,
226
+ "reasoning_output_tokens": 0
227
+ },
228
+ {
229
+ "cache_write_input_tokens": 0,
230
+ "cached_input_tokens": 30976,
231
+ "input_tokens": 31184,
232
+ "output_tokens": 288,
233
+ "reasoning_output_tokens": 174
234
+ },
235
+ {
236
+ "cache_write_input_tokens": 0,
237
+ "cached_input_tokens": 31360,
238
+ "input_tokens": 31984,
239
+ "output_tokens": 160,
240
+ "reasoning_output_tokens": 83
241
+ },
242
+ {
243
+ "cache_write_input_tokens": 0,
244
+ "cached_input_tokens": 32000,
245
+ "input_tokens": 32638,
246
+ "output_tokens": 56,
247
+ "reasoning_output_tokens": 21
248
+ },
249
+ {
250
+ "cache_write_input_tokens": 0,
251
+ "cached_input_tokens": 32512,
252
+ "input_tokens": 33487,
253
+ "output_tokens": 23,
254
+ "reasoning_output_tokens": 0
255
+ },
256
+ {
257
+ "cache_write_input_tokens": 0,
258
+ "cached_input_tokens": 33408,
259
+ "input_tokens": 33528,
260
+ "output_tokens": 23,
261
+ "reasoning_output_tokens": 0
262
+ },
263
+ {
264
+ "cache_write_input_tokens": 0,
265
+ "cached_input_tokens": 33408,
266
+ "input_tokens": 33721,
267
+ "output_tokens": 555,
268
+ "reasoning_output_tokens": 52
269
+ },
270
+ {
271
+ "cache_write_input_tokens": 0,
272
+ "cached_input_tokens": 34176,
273
+ "input_tokens": 34296,
274
+ "output_tokens": 40,
275
+ "reasoning_output_tokens": 0
276
+ },
277
+ {
278
+ "cache_write_input_tokens": 0,
279
+ "cached_input_tokens": 34176,
280
+ "input_tokens": 34670,
281
+ "output_tokens": 193,
282
+ "reasoning_output_tokens": 117
283
+ },
284
+ {
285
+ "cache_write_input_tokens": 0,
286
+ "cached_input_tokens": 34688,
287
+ "input_tokens": 35365,
288
+ "output_tokens": 47,
289
+ "reasoning_output_tokens": 13
290
+ },
291
+ {
292
+ "cache_write_input_tokens": 0,
293
+ "cached_input_tokens": 35200,
294
+ "input_tokens": 35474,
295
+ "output_tokens": 345,
296
+ "reasoning_output_tokens": 300
297
+ },
298
+ {
299
+ "cache_write_input_tokens": 0,
300
+ "cached_input_tokens": 35584,
301
+ "input_tokens": 36417,
302
+ "output_tokens": 453,
303
+ "reasoning_output_tokens": 373
304
+ },
305
+ {
306
+ "cache_write_input_tokens": 0,
307
+ "cached_input_tokens": 36736,
308
+ "input_tokens": 37369,
309
+ "output_tokens": 315,
310
+ "reasoning_output_tokens": 223
311
+ },
312
+ {
313
+ "cache_write_input_tokens": 0,
314
+ "cached_input_tokens": 37632,
315
+ "input_tokens": 38183,
316
+ "output_tokens": 59,
317
+ "reasoning_output_tokens": 28
318
+ },
319
+ {
320
+ "cache_write_input_tokens": 0,
321
+ "cached_input_tokens": 38144,
322
+ "input_tokens": 38304,
323
+ "output_tokens": 93,
324
+ "reasoning_output_tokens": 51
325
+ },
326
+ {
327
+ "cache_write_input_tokens": 0,
328
+ "cached_input_tokens": 38272,
329
+ "input_tokens": 38878,
330
+ "output_tokens": 201,
331
+ "reasoning_output_tokens": 80
332
+ },
333
+ {
334
+ "cache_write_input_tokens": 0,
335
+ "cached_input_tokens": 38912,
336
+ "input_tokens": 39581,
337
+ "output_tokens": 32,
338
+ "reasoning_output_tokens": 0
339
+ },
340
+ {
341
+ "cache_write_input_tokens": 0,
342
+ "cached_input_tokens": 39424,
343
+ "input_tokens": 40093,
344
+ "output_tokens": 128,
345
+ "reasoning_output_tokens": 44
346
+ },
347
+ {
348
+ "cache_write_input_tokens": 0,
349
+ "cached_input_tokens": 40064,
350
+ "input_tokens": 40714,
351
+ "output_tokens": 87,
352
+ "reasoning_output_tokens": 32
353
+ },
354
+ {
355
+ "cache_write_input_tokens": 0,
356
+ "cached_input_tokens": 40576,
357
+ "input_tokens": 41287,
358
+ "output_tokens": 105,
359
+ "reasoning_output_tokens": 0
360
+ },
361
+ {
362
+ "cache_write_input_tokens": 0,
363
+ "cached_input_tokens": 41216,
364
+ "input_tokens": 41902,
365
+ "output_tokens": 40,
366
+ "reasoning_output_tokens": 0
367
+ },
368
+ {
369
+ "cache_write_input_tokens": 0,
370
+ "cached_input_tokens": 41728,
371
+ "input_tokens": 42274,
372
+ "output_tokens": 111,
373
+ "reasoning_output_tokens": 36
374
+ },
375
+ {
376
+ "cache_write_input_tokens": 0,
377
+ "cached_input_tokens": 42240,
378
+ "input_tokens": 42883,
379
+ "output_tokens": 56,
380
+ "reasoning_output_tokens": 0
381
+ },
382
+ {
383
+ "cache_write_input_tokens": 0,
384
+ "cached_input_tokens": 42880,
385
+ "input_tokens": 43430,
386
+ "output_tokens": 54,
387
+ "reasoning_output_tokens": 0
388
+ },
389
+ {
390
+ "cache_write_input_tokens": 0,
391
+ "cached_input_tokens": 43392,
392
+ "input_tokens": 43967,
393
+ "output_tokens": 81,
394
+ "reasoning_output_tokens": 0
395
+ },
396
+ {
397
+ "cache_write_input_tokens": 0,
398
+ "cached_input_tokens": 43904,
399
+ "input_tokens": 44545,
400
+ "output_tokens": 66,
401
+ "reasoning_output_tokens": 32
402
+ },
403
+ {
404
+ "cache_write_input_tokens": 0,
405
+ "cached_input_tokens": 44416,
406
+ "input_tokens": 45087,
407
+ "output_tokens": 225,
408
+ "reasoning_output_tokens": 111
409
+ },
410
+ {
411
+ "cache_write_input_tokens": 0,
412
+ "cached_input_tokens": 45184,
413
+ "input_tokens": 45813,
414
+ "output_tokens": 101,
415
+ "reasoning_output_tokens": 29
416
+ },
417
+ {
418
+ "cache_write_input_tokens": 0,
419
+ "cached_input_tokens": 45696,
420
+ "input_tokens": 46415,
421
+ "output_tokens": 32,
422
+ "reasoning_output_tokens": 0
423
+ },
424
+ {
425
+ "cache_write_input_tokens": 0,
426
+ "cached_input_tokens": 46208,
427
+ "input_tokens": 46923,
428
+ "output_tokens": 112,
429
+ "reasoning_output_tokens": 54
430
+ },
431
+ {
432
+ "cache_write_input_tokens": 0,
433
+ "cached_input_tokens": 46976,
434
+ "input_tokens": 47169,
435
+ "output_tokens": 99,
436
+ "reasoning_output_tokens": 47
437
+ },
438
+ {
439
+ "cache_write_input_tokens": 0,
440
+ "cached_input_tokens": 47232,
441
+ "input_tokens": 47665,
442
+ "output_tokens": 745,
443
+ "reasoning_output_tokens": 154
444
+ },
445
+ {
446
+ "cache_write_input_tokens": 0,
447
+ "cached_input_tokens": 48256,
448
+ "input_tokens": 48433,
449
+ "output_tokens": 41,
450
+ "reasoning_output_tokens": 0
451
+ }
452
+ ],
453
+ "usage_source": "provider-raw"
454
+ },
455
+ "call_activity": {
456
+ "schema": "rlebench/call-activity/1",
457
+ "source": "Kinex native trace",
458
+ "model_tool_calls": 61,
459
+ "model_tool_calls_by_name": {
460
+ "ls": 1,
461
+ "read": 15,
462
+ "find": 1,
463
+ "exec": 32,
464
+ "devices": 2,
465
+ "write": 3,
466
+ "edit": 1,
467
+ "device": 4,
468
+ "device_stop": 2
469
+ },
470
+ "python_instrumented_model_tool_calls": null,
471
+ "python_tool_invocations": null,
472
+ "nested_python_tool_invocations": null,
473
+ "python_tool_invocations_by_origin": null,
474
+ "python_device_rpc_attempts": null,
475
+ "python_device_rpc_attempts_by_action": null,
476
+ "python_device_rpc_errors": null,
477
+ "python_scope": "Recorded instrumented Python invocations only; other paths and automatic fatal stops are not counted."
478
+ },
479
+ "model_response_count": 54,
480
+ "usage_source": "Harbor session metrics",
481
+ "execution": {
482
+ "status": "finished",
483
+ "reason": null
484
+ },
485
+ "native": {
486
+ "evidence_valid": true,
487
+ "success": true,
488
+ "termination": "success",
489
+ "steps": 621
490
+ },
491
+ "failure_attribution": {
492
+ "classification": "native_success",
493
+ "physical_cause": null,
494
+ "physical_cause_evidence": null
495
+ },
496
+ "exception": null
497
+ }
experiments/robotwin-native/kinex-1/episode.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "episode": 1,
3
+ "initialization": {
4
+ "expert_plan_success": false,
5
+ "requested_seed": 0
6
+ },
7
+ "native_reward": 1.0,
8
+ "planned_episodes": 1,
9
+ "reason": "success",
10
+ "simulation_time_s": 24.84,
11
+ "step_limit": 800,
12
+ "steps": 621,
13
+ "success": true,
14
+ "valid": true
15
+ }
experiments/robotwin-native/kinex-1/evidence.tar.gz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:66000d8bb81d51f2a4baecf42b7fcaaf237e23a1a544fe68e3c4e0f6073a9042
3
+ size 5690824
experiments/robotwin-native/kinex-1/instruction.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "binding_paths": {},
3
+ "bindings": {},
4
+ "context": {},
5
+ "instruction": "use the left arm to grasp the red block on the table, hand it over to the right arm, and set it down in the center of the blue pad. Release it with the right gripper open",
6
+ "native_instruction": "",
7
+ "policy": "modified",
8
+ "prompt": "# Task: robotwin/handover-block-aloha-agilex\n\nuse the left arm to grasp the red block on the table, hand it over to the right arm, and set it down in the center of the blue pad. Release it with the right gripper open\n\n## Work to do\n\nSolve the task efficiently using the running rosbridge environment. Follow the\ntask instruction above, including its completion requirements. Read `harness.md`\nand the device profiles for public interfaces, units and calibrated frames.\nReuse your saved tools, skills, and memos.\n\n## Limits\n\nYou have 800 native action frames. The execution mode is `stepped`: physics is\nfrozen while idle. Use `world.wait` with `duration_s` to advance the native clock.\nJoint targets are ROS JointTrajectory messages; gripper targets use the units\ndeclared in their profiles. Reads and command publication do not advance physics.\nWaits are sequential; concurrent waits are rejected. A command or elapsed time\ndoes not prove arrival; inspect measured feedback. Device stop holds the device\nwithout ending the episode. The safety timeout is 28800 seconds.\nDo not launch or reset a simulation.\n\n## Completion\n\nReturn when finished. The owner finalizes the episode and reads its native\nverdict. An interrupted or rejected wait does not establish task success. The\nnative step limit stops physics even if your requested wait is longer.\n\n## Saved resources\n\nMaintain reusable code in `tools/`, skills in `skills/`, and concise working notes\nin `memos/`. Improve existing files in place as you learn, replacing outdated\ncontent. Create a new file only when no existing resource fits; do not create\nper-attempt reflections or numbered copies. These files persist across fresh\nconversations. Update them before finishing your response.\n\n## Robot and device access\n\nUse Kinex's ROS device profiles in `.kinex/devices/` and the robot\nspecification in `.kinex/workspace.yaml`. The adapter refreshes them each episode.\nUse `kinex.communication` for typed topic publication, subscriptions and service\nrequests. Inspect operation schemas before acting. Load the provided devices;\nthey borrow the running simulator. Stop holds motion; it does not reset the world.\nImprove existing tools, skills, and memos in place.\n\n",
9
+ "prompt_template": "<!-- rlebench:resolved-episode-instruction -->\n# Task: robotwin/handover-block-aloha-agilex\n\nuse the left arm to grasp the red block on the table, hand it over to the right arm, and set it down in the center of the blue pad. Release it with the right gripper open\n\n## Work to do\n\nSolve the task efficiently using the running rosbridge environment. Follow the\ntask instruction above, including its completion requirements. Read `harness.md`\nand the device profiles for public interfaces, units and calibrated frames.\nReuse your saved tools, skills, and memos.\n\n## Limits\n\nYou have 800 native action frames. The execution mode is `stepped`: physics is\nfrozen while idle. Use `world.wait` with `duration_s` to advance the native clock.\nJoint targets are ROS JointTrajectory messages; gripper targets use the units\ndeclared in their profiles. Reads and command publication do not advance physics.\nWaits are sequential; concurrent waits are rejected. A command or elapsed time\ndoes not prove arrival; inspect measured feedback. Device stop holds the device\nwithout ending the episode. The safety timeout is 28800 seconds.\nDo not launch or reset a simulation.\n\n## Completion\n\nReturn when finished. The owner finalizes the episode and reads its native\nverdict. An interrupted or rejected wait does not establish task success. The\nnative step limit stops physics even if your requested wait is longer.\n\n## Saved resources\n\nMaintain reusable code in `tools/`, skills in `skills/`, and concise working notes\nin `memos/`. Improve existing files in place as you learn, replacing outdated\ncontent. Create a new file only when no existing resource fits; do not create\nper-attempt reflections or numbered copies. These files persist across fresh\nconversations. Update them before finishing your response.\n\n## Robot and device access\n\nUse Kinex's ROS device profiles in `.kinex/devices/` and the robot\nspecification in `.kinex/workspace.yaml`. The adapter refreshes them each episode.\nUse `kinex.communication` for typed topic publication, subscriptions and service\nrequests. Inspect operation schemas before acting. Load the provided devices;\nthey borrow the running simulator. Stop holds motion; it does not reset the world.\nImprove existing tools, skills, and memos in place.\n\n",
10
+ "schema": "rlebench/episode-instruction/3",
11
+ "source": "catalog",
12
+ "task_id": "robotwin/handover-block-aloha-agilex",
13
+ "template": "use the left arm to grasp the red block on the table, hand it over to the right arm, and set it down in the center of the blue pad. Release it with the right gripper open"
14
+ }
experiments/robotwin-native/kinex-1/media.json ADDED
@@ -0,0 +1,101 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "native_frames": 250,
3
+ "published_frames": 250,
4
+ "pts_verified": true,
5
+ "source_start_pts": 0.2,
6
+ "published_start_pts": 0.0,
7
+ "pts_offset": -0.2,
8
+ "terminal_frame_retained": true,
9
+ "views": [
10
+ "third_person",
11
+ "left_wrist",
12
+ "right_wrist"
13
+ ],
14
+ "fps": "10/1",
15
+ "native_recording": {
16
+ "schema": "rosbridge/recording/1",
17
+ "run_id": "7b75d3e8384a",
18
+ "task": "robotwin/handover-block-aloha-agilex",
19
+ "state": "complete",
20
+ "views": {
21
+ "third_person": {
22
+ "received": 250,
23
+ "dropped": 0,
24
+ "segments": [
25
+ {
26
+ "file": "third_person.mp4",
27
+ "frames": 250,
28
+ "width": 960,
29
+ "height": 720,
30
+ "fps": 10.0,
31
+ "first_stamp_ns": 0,
32
+ "last_stamp_ns": 24840000000
33
+ }
34
+ ],
35
+ "error": null
36
+ },
37
+ "left_wrist": {
38
+ "received": 250,
39
+ "dropped": 0,
40
+ "segments": [
41
+ {
42
+ "file": "left_wrist.mp4",
43
+ "frames": 250,
44
+ "width": 960,
45
+ "height": 720,
46
+ "fps": 10.0,
47
+ "first_stamp_ns": 0,
48
+ "last_stamp_ns": 24840000000
49
+ }
50
+ ],
51
+ "error": null
52
+ },
53
+ "right_wrist": {
54
+ "received": 250,
55
+ "dropped": 0,
56
+ "segments": [
57
+ {
58
+ "file": "right_wrist.mp4",
59
+ "frames": 250,
60
+ "width": 960,
61
+ "height": 720,
62
+ "fps": 10.0,
63
+ "first_stamp_ns": 0,
64
+ "last_stamp_ns": 24840000000
65
+ }
66
+ ],
67
+ "error": null
68
+ }
69
+ },
70
+ "type": "mp4",
71
+ "spectator_views": [
72
+ "third_person",
73
+ "left_wrist",
74
+ "right_wrist"
75
+ ],
76
+ "fps": 10.0,
77
+ "topics": [],
78
+ "error": null,
79
+ "bag_messages": 0
80
+ },
81
+ "all_frame_visual_checks": {
82
+ "third_person": {
83
+ "frames_compared": 250,
84
+ "mean_ssim": 0.9981849039999999,
85
+ "minimum_ssim": 0.996967,
86
+ "label_area_excluded": true
87
+ },
88
+ "left_wrist": {
89
+ "frames_compared": 250,
90
+ "mean_ssim": 0.998478984,
91
+ "minimum_ssim": 0.996485,
92
+ "label_area_excluded": true
93
+ },
94
+ "right_wrist": {
95
+ "frames_compared": 250,
96
+ "mean_ssim": 0.998197364,
97
+ "minimum_ssim": 0.995219,
98
+ "label_area_excluded": true
99
+ }
100
+ }
101
+ }
experiments/robotwin-native/kinex-1/observations/01d7a694-1.png ADDED
experiments/robotwin-native/kinex-1/observations/0bd66e5e-1.png ADDED
experiments/robotwin-native/kinex-1/observations/0d7d46d8-1.png ADDED
experiments/robotwin-native/kinex-1/observations/13f789a0-1.png ADDED
experiments/robotwin-native/kinex-1/observations/1953a6e9-1.png ADDED
experiments/robotwin-native/kinex-1/observations/245fef43-1.png ADDED
experiments/robotwin-native/kinex-1/observations/3d9b5793-1.png ADDED
experiments/robotwin-native/kinex-1/observations/451da7eb-1.png ADDED
experiments/robotwin-native/kinex-1/observations/4a6cdfeb-1.png ADDED
experiments/robotwin-native/kinex-1/observations/4f33bd50-1.png ADDED
experiments/robotwin-native/kinex-1/observations/5c5beb92-1.png ADDED
experiments/robotwin-native/kinex-1/observations/5df3e384-1.png ADDED
experiments/robotwin-native/kinex-1/observations/5ffc47a8-1.png ADDED
experiments/robotwin-native/kinex-1/observations/797caa5e-1.png ADDED
experiments/robotwin-native/kinex-1/observations/86db988c-1.png ADDED
experiments/robotwin-native/kinex-1/observations/86db988c-2.png ADDED
experiments/robotwin-native/kinex-1/observations/a51b5070-1.png ADDED
experiments/robotwin-native/kinex-1/observations/a8adce0d-1.png ADDED
experiments/robotwin-native/kinex-1/observations/af212c1a-1.png ADDED
experiments/robotwin-native/kinex-1/observations/b5d4c986-1.png ADDED
experiments/robotwin-native/kinex-1/observations/b833ee1b-1.png ADDED
experiments/robotwin-native/kinex-1/observations/bab02154-1.png ADDED
experiments/robotwin-native/kinex-1/observations/bf1286a6-1.png ADDED
experiments/robotwin-native/kinex-1/observations/c1d037d5-1.png ADDED
experiments/robotwin-native/kinex-1/observations/c3226fa8-1.png ADDED
experiments/robotwin-native/kinex-1/observations/db1b4a52-1.png ADDED
experiments/robotwin-native/kinex-1/observations/e7cd2d3f-1.png ADDED
experiments/robotwin-native/kinex-1/observations/e84d68b9-1.png ADDED
experiments/robotwin-native/kinex-1/observations/f0765732-1.png ADDED
experiments/robotwin-native/kinex-1/observations/f9d2862d-1.png ADDED
experiments/robotwin-native/kinex-1/policy-calls.json ADDED
@@ -0,0 +1 @@
 
 
1
+ []
experiments/robotwin-native/kinex-1/poster.jpg ADDED
experiments/robotwin-native/kinex-1/provider-requests.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
experiments/robotwin-native/kinex-1/session.json ADDED
The diff for this file is too large to render. See raw diff
 
experiments/robotwin-native/kinex-1/session.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
experiments/robotwin-native/kinex-1/terminal.jpg ADDED
experiments/robotwin-native/kinex-1/tool-calls.json ADDED
The diff for this file is too large to render. See raw diff
 
experiments/robotwin-native/kinex-1/trace.json ADDED
The diff for this file is too large to render. See raw diff