syvb commited on
Commit
8769737
·
verified ·
1 Parent(s): 6e7f501

Add noise run log

Browse files
Files changed (1) hide show
  1. logs/noise.log +134 -0
logs/noise.log ADDED
@@ -0,0 +1,134 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ == 17:49:13 install
2
+ WARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager, possibly rendering your system unusable. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv. Use the --root-user-action option if you know what you are doing and want to suppress this warning.
3
+ == 17:49:35 download data
4
+ Hint: The `hf-cli` skill is not installed. Run `hf skills add -g` to teach your AI agents how to use the `hf` CLI.
5
+ Hint: The `hf-cli` skill is not installed. Run `hf skills add -g` to teach your AI agents how to use the `hf` CLI.
6
+
7
+ Warning: You are sending unauthenticated requests to the HF Hub. Please set a HF_TOKEN to enable higher rate limits and faster downloads.
8
+
9
+ path: /workspace/act-experiments/data/dl/sample/10BT/000_00000.parquet
10
+ == 17:51:19 download means
11
+
12
+ path: /workspace/act-experiments/out/means_14B.pt
13
+ == 17:55:41 test
14
+ Warning: You are sending unauthenticated requests to the HF Hub. Please set a HF_TOKEN to enable higher rate limits and faster downloads.
15
+
16
+ replace at site -1: max relative error resid 1.7e-06
17
+ replace at site -1: max relative error attn 2.9e-06
18
+ replace at site -1: max relative error mlp 3.2e-06
19
+ replace at site -1: max relative error logits 1.3e-06
20
+ replace at site 0: max relative error resid 2.4e-06
21
+ replace at site 0: max relative error attn 3.9e-06
22
+ replace at site 0: max relative error mlp 3.9e-06
23
+ replace at site 0: max relative error logits 4.2e-06
24
+ replace at site 1: max relative error resid 2.2e-06
25
+ replace at site 1: max relative error attn 4.3e-06
26
+ replace at site 1: max relative error mlp 3.0e-06
27
+ replace at site 1: max relative error logits 1.4e-06
28
+ replace at site 14: max relative error resid 5.3e-06
29
+ replace at site 14: max relative error attn 7.8e-06
30
+ replace at site 14: max relative error mlp 8.6e-06
31
+ replace at site 14: max relative error logits 6.0e-06
32
+ replace at site 28: max relative error resid 1.7e-06
33
+ replace at site 28: max relative error attn 2.9e-06
34
+ replace at site 28: max relative error mlp 3.2e-06
35
+ replace at site 28: max relative error logits 3.4e-07
36
+ add at site -1: max relative error resid 1.7e-06
37
+ add at site -1: max relative error attn 2.9e-06
38
+ add at site -1: max relative error mlp 3.2e-06
39
+ add at site -1: max relative error logits 1.3e-06
40
+ add at site 0: max relative error resid 2.5e-06
41
+ add at site 0: max relative error attn 4.7e-06
42
+ add at site 0: max relative error mlp 3.8e-06
43
+ add at site 0: max relative error logits 3.1e-06
44
+ add at site 1: max relative error resid 1.7e-06
45
+ add at site 1: max relative error attn 3.5e-06
46
+ add at site 1: max relative error mlp 3.3e-06
47
+ add at site 1: max relative error logits 1.2e-06
48
+ add at site 14: max relative error resid 1.4e-06
49
+ add at site 14: max relative error attn 2.5e-06
50
+ add at site 14: max relative error mlp 2.9e-06
51
+ add at site 14: max relative error logits 1.4e-06
52
+ add at site 28: max relative error resid 1.7e-06
53
+ add at site 28: max relative error attn 2.9e-06
54
+ add at site 28: max relative error mlp 3.2e-06
55
+ add at site 28: max relative error logits 1.3e-06
56
+ OK
57
+
58
+ == 17:56:24 noise 14B
59
+ Warning: You are sending unauthenticated requests to the HF Hub. Please set a HF_TOKEN to enable higher rate limits and faster downloads.
60
+
61
+ 100/3000 samples
62
+ 200/3000 samples
63
+ 300/3000 samples
64
+ 400/3000 samples
65
+ 500/3000 samples
66
+ 600/3000 samples
67
+ 700/3000 samples
68
+ 800/3000 samples
69
+ 900/3000 samples
70
+ 1000/3000 samples
71
+ 1100/3000 samples
72
+ 1200/3000 samples
73
+ 1300/3000 samples
74
+ 1400/3000 samples
75
+ 1500/3000 samples
76
+ 1600/3000 samples
77
+ 1700/3000 samples
78
+ 1800/3000 samples
79
+ 1900/3000 samples
80
+ 2000/3000 samples
81
+ 2100/3000 samples
82
+ 2200/3000 samples
83
+ 2300/3000 samples
84
+ 2400/3000 samples
85
+ 2500/3000 samples
86
+ 2600/3000 samples
87
+ 2700/3000 samples
88
+ 2800/3000 samples
89
+ 2900/3000 samples
90
+ 3000/3000 samples
91
+ wandb: [wandb.login()] Loaded credentials for https://api.wandb.ai from WANDB_API_KEY.
92
+ wandb: Currently logged in as: syvb (octahedral-systems) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
93
+ wandb: setting up run vb41ya95
94
+ wandb: Tracking run with wandb version 0.30.0
95
+ wandb: Run data is saved locally in /workspace/act-experiments/wandb/run-20261015_190022-vb41ya95
96
+ wandb: Run `wandb offline` to turn off syncing.
97
+ wandb: Syncing run Qwen3-14B-Base
98
+ wandb: ⭐️ View project at https://wandb.ai/octahedral-systems/qwen3-self-correction-noise
99
+ wandb: 🚀 View run at https://wandb.ai/octahedral-systems/qwen3-self-correction-noise/runs/vb41ya95
100
+ wandb: updating run metadata
101
+ wandb: uploading requirements.txt; uploading wandb-summary.json; uploading config.yaml; uploading wandb-metadata.json
102
+ wandb:
103
+ wandb: Run history:
104
+ wandb: add_context_x0.25/disruption ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂▂▂▂▂▂▂▃▃▃▄▅▅▆▆▇████
105
+ wandb: add_context_x0.25/dloss ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂▂▂▂▂▂▃▃▃▃▃▃▃▃▄█
106
+ wandb: add_context_x0.25/kl ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂▂▂▂▂▂▃▃▃▃▃▃▃▃█
107
+ wandb: add_context_x0.25/retained ▁▁▁▁▁▁▁▁▁▂▂▂▂▂▂▃▃▃▄▄▄▅▅▆▆▇▇███████▇▇▇▇▇
108
+ wandb: add_context_x1/disruption ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂▂▂▂▂▂▂▂▃▃▃▄▄▄▅▅▆▆▆▇▇▇▇█
109
+ wandb: add_context_x1/dloss ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂▂▂▂▂▂▂▃▃▃▃▄▄▄▄▄▄▄▅█
110
+ wandb: add_context_x1/kl ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂▂▂▂▂▂▂▃▃▃▃▄▄▄▄▄▄▄▅█
111
+ wandb: add_context_x1/retained ▂▃▁▃▃▃▃▃▃▃▃▃▃▃▃▄▄▄▅▅▆▆▆▇▇██████▇▇▇▇▇▇▇▇
112
+ wandb: add_context_x16/disruption ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂▂▂▂▂▃█
113
+ wandb: add_context_x16/dloss ▁▂▂▂▂▂▂▂▂▃▃▃▃▃▃▃▃▃▃▃▄▄▄▄▄▄▄▄▅▅▅▆▆▇▇▇███▆
114
+ wandb: +33 ...
115
+ wandb:
116
+ wandb: Run summary:
117
+ wandb: add_context_x0.25/disruption 0.06494
118
+ wandb: add_context_x0.25/dloss 0.11054
119
+ wandb: add_context_x0.25/kl 0.11625
120
+ wandb: add_context_x0.25/retained 1.0
121
+ wandb: add_context_x1/disruption 1.03903
122
+ wandb: add_context_x1/dloss 2.17651
123
+ wandb: add_context_x1/kl 2.19894
124
+ wandb: add_context_x1/retained 1.0
125
+ wandb: add_context_x16/disruption 265.99063
126
+ wandb: add_context_x16/dloss 18.12283
127
+ wandb: +34 ...
128
+ wandb:
129
+ wandb: 🚀 View run Qwen3-14B-Base at: https://wandb.ai/octahedral-systems/qwen3-self-correction-noise/runs/vb41ya95
130
+ wandb: ⭐️ View project at: https://wandb.ai/octahedral-systems/qwen3-self-correction-noise
131
+ wandb: Synced 4 W&B file(s), 0 media file(s), 0 artifact file(s) and 0 other file(s)
132
+ wandb: Find logs at: ./wandb/run-20261015_190022-vb41ya95/logs
133
+ == 19:00:29 plot
134
+ == 19:00:36 done