kacperwikiel commited on
Commit
79fa441
·
verified ·
1 Parent(s): 5db92c8

Update experiment measurements and status

Browse files
reports/configs/control-1337.toml ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [model]
2
+ vocab = 12288
3
+ n_layer = 19
4
+ n_embd = 768
5
+ n_head = 12
6
+ block = 1024
7
+ ffn_mult = 2.8125
8
+ [train]
9
+ seed = 1337
10
+ micro_batch = 64
11
+ global_batch = 512
12
+ steps = 1000
13
+ warmup = 50
14
+ lr = 0.0001
15
+ min_lr = 0.00001
16
+ muon_lr = 0.0033333333333333335
17
+ checkpoint_every = 500
18
+ validate_every = 500
19
+ log_every = 20
20
+ compile = true
21
+ [data]
22
+ train = "/data/ablation/mixtures/control/train.bin"
23
+ validation = "/data/final/val.bin"
24
+ sha256 = "b7f7f59462cac9092bd301e742008413ddfbbd81cb1b2e901bea0d9f9e275ecf"
reports/eval/fineweb_edu-1337-1000.json ADDED
The diff for this file is too large to render. See raw diff
 
reports/runs/control-1337/resolved.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model": {
3
+ "vocab": 12288,
4
+ "n_layer": 19,
5
+ "n_embd": 768,
6
+ "n_head": 12,
7
+ "block": 1024,
8
+ "ffn_mult": 2.8125,
9
+ "norm": "rmsnorm",
10
+ "norm_eps": 1e-06,
11
+ "pos": "rope",
12
+ "rope_theta": 100000.0,
13
+ "ffn": "swiglu",
14
+ "value_residual": true,
15
+ "qk_norm": true,
16
+ "logit_cap": 0.0,
17
+ "z_loss": 0.0
18
+ },
19
+ "train": {
20
+ "seed": 1337,
21
+ "micro_batch": 64,
22
+ "global_batch": 512,
23
+ "steps": 1000,
24
+ "warmup": 50,
25
+ "lr": 0.0001,
26
+ "min_lr": 1e-05,
27
+ "muon_lr": 0.0033333333333333335,
28
+ "checkpoint_every": 500,
29
+ "validate_every": 500,
30
+ "log_every": 20,
31
+ "compile": true
32
+ },
33
+ "data": {
34
+ "train": "/data/ablation/mixtures/control/train.bin",
35
+ "validation": "/data/final/val.bin",
36
+ "sha256": "b7f7f59462cac9092bd301e742008413ddfbbd81cb1b2e901bea0d9f9e275ecf"
37
+ }
38
+ }
reports/status.json CHANGED
@@ -1,6 +1,8 @@
1
  {
2
- "stage": "development_evaluation",
3
- "utc": "2026-09-30T21:43:11.686250+00:00",
4
- "remaining_seconds": 37542.313722,
5
- "checkpoint": "/data/ablation/runs/fineweb_edu-1337/checkpoint-000001000.pt"
 
 
6
  }
 
1
  {
2
+ "stage": "training",
3
+ "utc": "2026-09-30T21:45:23.042848+00:00",
4
+ "remaining_seconds": 37410.957035,
5
+ "source": "control",
6
+ "seed": 1337,
7
+ "adaptation_tokens": 524288000
8
  }