agatha_v2_coco256_agatha_s0: records to step 100000
Browse files
coco256/agatha_v2_coco256_agatha_s0/agatha_v2_coco256_agatha_s0_evals.jsonl
CHANGED
|
@@ -152,3 +152,4 @@
|
|
| 152 |
{"run": "agatha_v2_coco256_agatha_s0", "step": 100000, "hook": "fid", "ok": true, "seconds": 173.03, "result": {"prompts": 2000, "raw_g1": {"fid": 67.06709058152757, "n_real": 40504, "n_fake": 2000, "kid_mean": 0.04017759672339551, "kid_std": 0.0022481554245341027, "subset_size_used": 1000, "subsets": 100}, "raw_g3": {"fid": 35.885688503293125, "n_real": 40504, "n_fake": 2000, "kid_mean": 0.004596426764285209, "kid_std": 0.00027717597859134826, "subset_size_used": 1000, "subsets": 100}, "ema_g1": {"fid": 66.74275072823389, "n_real": 40504, "n_fake": 2000, "kid_mean": 0.03987953198437659, "kid_std": 0.0021669125461453464, "subset_size_used": 1000, "subsets": 100}, "ema_g3": {"fid": 34.33272820368808, "n_real": 40504, "n_fake": 2000, "kid_mean": 0.004278491686441184, "kid_std": 0.0002631027457333794, "subset_size_used": 1000, "subsets": 100}}}
|
| 153 |
{"run": "agatha_v2_coco256_agatha_s0", "step": 100000, "hook": "clip", "ok": true, "seconds": 93.1, "result": {"items": 500, "raw_g1": {"clip_score": {"mean": 18.859185100378376, "se": 0.1720976747536777, "n": 500}, "grounding": {"mean": 0.05764315659746718, "se": 0.002420565985569894, "n": 500, "own_cos": 0.6094668357188114, "shuffled_cos": 0.5518236791213443}}, "raw_g3": {"clip_score": {"mean": 22.472661529536513, "se": 0.17254688190966583, "n": 500}, "grounding": {"mean": 0.12684449857455463, "se": 0.0036984928807680304, "n": 500, "own_cos": 0.6705515249469666, "shuffled_cos": 0.543707026372412}}, "ema_g1": {"clip_score": {"mean": 18.78178263357732, "se": 0.164180003169551, "n": 500}, "grounding": {"mean": 0.05941550538668376, "se": 0.002380165330532102, "n": 500, "own_cos": 0.6104982853900163, "shuffled_cos": 0.5510827800033326}}, "ema_g3": {"clip_score": {"mean": 22.564667563482494, "se": 0.1732170623614698, "n": 500}, "grounding": {"mean": 0.1264531956569199, "se": 0.0037610935704458963, "n": 500, "own_cos": 0.6714111774395024, "shuffled_cos": 0.5449579817825826}}}}
|
| 154 |
{"run": "agatha_v2_coco256_agatha_s0", "step": 100000, "hook": "checkpoint", "ok": true, "seconds": 8.12, "result": {"file": "agatha_v2_coco256_agatha_s0_state_step100000.pt", "bytes": 672811440, "pushed": true}}
|
|
|
|
|
|
| 152 |
{"run": "agatha_v2_coco256_agatha_s0", "step": 100000, "hook": "fid", "ok": true, "seconds": 173.03, "result": {"prompts": 2000, "raw_g1": {"fid": 67.06709058152757, "n_real": 40504, "n_fake": 2000, "kid_mean": 0.04017759672339551, "kid_std": 0.0022481554245341027, "subset_size_used": 1000, "subsets": 100}, "raw_g3": {"fid": 35.885688503293125, "n_real": 40504, "n_fake": 2000, "kid_mean": 0.004596426764285209, "kid_std": 0.00027717597859134826, "subset_size_used": 1000, "subsets": 100}, "ema_g1": {"fid": 66.74275072823389, "n_real": 40504, "n_fake": 2000, "kid_mean": 0.03987953198437659, "kid_std": 0.0021669125461453464, "subset_size_used": 1000, "subsets": 100}, "ema_g3": {"fid": 34.33272820368808, "n_real": 40504, "n_fake": 2000, "kid_mean": 0.004278491686441184, "kid_std": 0.0002631027457333794, "subset_size_used": 1000, "subsets": 100}}}
|
| 153 |
{"run": "agatha_v2_coco256_agatha_s0", "step": 100000, "hook": "clip", "ok": true, "seconds": 93.1, "result": {"items": 500, "raw_g1": {"clip_score": {"mean": 18.859185100378376, "se": 0.1720976747536777, "n": 500}, "grounding": {"mean": 0.05764315659746718, "se": 0.002420565985569894, "n": 500, "own_cos": 0.6094668357188114, "shuffled_cos": 0.5518236791213443}}, "raw_g3": {"clip_score": {"mean": 22.472661529536513, "se": 0.17254688190966583, "n": 500}, "grounding": {"mean": 0.12684449857455463, "se": 0.0036984928807680304, "n": 500, "own_cos": 0.6705515249469666, "shuffled_cos": 0.543707026372412}}, "ema_g1": {"clip_score": {"mean": 18.78178263357732, "se": 0.164180003169551, "n": 500}, "grounding": {"mean": 0.05941550538668376, "se": 0.002380165330532102, "n": 500, "own_cos": 0.6104982853900163, "shuffled_cos": 0.5510827800033326}}, "ema_g3": {"clip_score": {"mean": 22.564667563482494, "se": 0.1732170623614698, "n": 500}, "grounding": {"mean": 0.1264531956569199, "se": 0.0037610935704458963, "n": 500, "own_cos": 0.6714111774395024, "shuffled_cos": 0.5449579817825826}}}}
|
| 154 |
{"run": "agatha_v2_coco256_agatha_s0", "step": 100000, "hook": "checkpoint", "ok": true, "seconds": 8.12, "result": {"file": "agatha_v2_coco256_agatha_s0_state_step100000.pt", "bytes": 672811440, "pushed": true}}
|
| 155 |
+
{"run": "agatha_v2_coco256_agatha_s0", "step": 100000, "hook": "milestone", "ok": true, "seconds": 2.22, "result": {"files": ["agatha_v2_coco256_agatha_s0_step100000_raw.pt", "agatha_v2_coco256_agatha_s0_step100000_ema.pt", "agatha_v2_coco256_agatha_s0_step100000_raw_g1.png", "agatha_v2_coco256_agatha_s0_step100000_raw_g3.png", "agatha_v2_coco256_agatha_s0_step100000_ema_g1.png", "agatha_v2_coco256_agatha_s0_step100000_ema_g3.png", "agatha_v2_coco256_agatha_s0_evals.jsonl", "agatha_v2_coco256_agatha_s0_train.log"], "pushed": true}}
|
coco256/agatha_v2_coco256_agatha_s0/agatha_v2_coco256_agatha_s0_summary_step100000.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"run": "agatha_v2_coco256_agatha_s0",
|
| 3 |
+
"start_step": 20000,
|
| 4 |
+
"step": 100000,
|
| 5 |
+
"end_step": 100000,
|
| 6 |
+
"last_mean_loss": 0.9782004952430725,
|
| 7 |
+
"seconds": 16569.6,
|
| 8 |
+
"stopped": "",
|
| 9 |
+
"arm": {
|
| 10 |
+
"arm": "agatha"
|
| 11 |
+
},
|
| 12 |
+
"peak_memory_gb": 21.17,
|
| 13 |
+
"hook_errors": [],
|
| 14 |
+
"device": "NVIDIA RTX PRO 6000 Blackwell Server Edition",
|
| 15 |
+
"torch": "2.14.1+cu130",
|
| 16 |
+
"recipe": {
|
| 17 |
+
"lr": 0.0001,
|
| 18 |
+
"betas": [
|
| 19 |
+
0.9,
|
| 20 |
+
0.999
|
| 21 |
+
],
|
| 22 |
+
"eps": 1e-08,
|
| 23 |
+
"batch_size": 256,
|
| 24 |
+
"steps": 100000,
|
| 25 |
+
"equalize_tower_gradients": true,
|
| 26 |
+
"bf16_autocast": true,
|
| 27 |
+
"windows_memory_fraction": 0.73,
|
| 28 |
+
"log_every": 100,
|
| 29 |
+
"checkpoint_every": 2000,
|
| 30 |
+
"out_dir": "/content/coco256/runs/agatha_v2_coco256_agatha_s0",
|
| 31 |
+
"seed": 0,
|
| 32 |
+
"num_workers": 8,
|
| 33 |
+
"ema_decay": 0.9999
|
| 34 |
+
}
|
| 35 |
+
}
|
coco256/agatha_v2_coco256_agatha_s0/agatha_v2_coco256_agatha_s0_train.log
CHANGED
|
@@ -1169,3 +1169,4 @@ arm agatha: {"arm": "agatha"}; 42,023,588 trained parameters; start step 20000
|
|
| 1169 |
[hook fid] step 100000 ok in 173.03 s: {"prompts": 2000, "raw_g1": {"fid": 67.06709058152757, "n_real": 40504, "n_fake": 2000, "kid_mean": 0.04017759672339551, "kid_std": 0.0022481554245341027, "subset_size_used": 1000, "subsets": 100}, "raw_g3": {"fid": 35.885688503293125, "n_real": 40504, "n_fake": 2000, "kid_mean": 0.004596426764285209, "kid_std": 0.00027717597859134826, "subset_size_used": 1000, "subsets": 100}, "ema_g1": {"fid": 66.74275072823389, "n_real": 40504, "n_fake": 2000, "kid_mean": 0.03987953198437659, "kid_std": 0.0021669125461453464, "subset_size_used": 1000, "subsets": 100}, "ema_g3": {"fid": 34.33272820368808, "n ...
|
| 1170 |
[hook clip] step 100000 ok in 93.1 s: {"items": 500, "raw_g1": {"clip_score": {"mean": 18.859185100378376, "se": 0.1720976747536777, "n": 500}, "grounding": {"mean": 0.05764315659746718, "se": 0.002420565985569894, "n": 500, "own_cos": 0.6094668357188114, "shuffled_cos": 0.5518236791213443}}, "raw_g3": {"clip_score": {"mean": 22.472661529536513, "se": 0.17254688190966583, "n": 500}, "grounding": {"mean": 0.12684449857455463, "se": 0.0036984928807680304, "n": 500, "own_cos": 0.6705515249469666, "shuffled_cos": 0.543707026372412}}, "ema_g1": {"clip_score": {"mean": 18.78178263357732, "se": 0.164180003169551, "n": 500}, "grounding": ...
|
| 1171 |
[hook checkpoint] step 100000 ok in 8.12 s: {"file": "agatha_v2_coco256_agatha_s0_state_step100000.pt", "bytes": 672811440, "pushed": true}
|
|
|
|
|
|
| 1169 |
[hook fid] step 100000 ok in 173.03 s: {"prompts": 2000, "raw_g1": {"fid": 67.06709058152757, "n_real": 40504, "n_fake": 2000, "kid_mean": 0.04017759672339551, "kid_std": 0.0022481554245341027, "subset_size_used": 1000, "subsets": 100}, "raw_g3": {"fid": 35.885688503293125, "n_real": 40504, "n_fake": 2000, "kid_mean": 0.004596426764285209, "kid_std": 0.00027717597859134826, "subset_size_used": 1000, "subsets": 100}, "ema_g1": {"fid": 66.74275072823389, "n_real": 40504, "n_fake": 2000, "kid_mean": 0.03987953198437659, "kid_std": 0.0021669125461453464, "subset_size_used": 1000, "subsets": 100}, "ema_g3": {"fid": 34.33272820368808, "n ...
|
| 1170 |
[hook clip] step 100000 ok in 93.1 s: {"items": 500, "raw_g1": {"clip_score": {"mean": 18.859185100378376, "se": 0.1720976747536777, "n": 500}, "grounding": {"mean": 0.05764315659746718, "se": 0.002420565985569894, "n": 500, "own_cos": 0.6094668357188114, "shuffled_cos": 0.5518236791213443}}, "raw_g3": {"clip_score": {"mean": 22.472661529536513, "se": 0.17254688190966583, "n": 500}, "grounding": {"mean": 0.12684449857455463, "se": 0.0036984928807680304, "n": 500, "own_cos": 0.6705515249469666, "shuffled_cos": 0.543707026372412}}, "ema_g1": {"clip_score": {"mean": 18.78178263357732, "se": 0.164180003169551, "n": 500}, "grounding": ...
|
| 1171 |
[hook checkpoint] step 100000 ok in 8.12 s: {"file": "agatha_v2_coco256_agatha_s0_state_step100000.pt", "bytes": 672811440, "pushed": true}
|
| 1172 |
+
[hook milestone] step 100000 ok in 2.22 s: {"files": ["agatha_v2_coco256_agatha_s0_step100000_raw.pt", "agatha_v2_coco256_agatha_s0_step100000_ema.pt", "agatha_v2_coco256_agatha_s0_step100000_raw_g1.png", "agatha_v2_coco256_agatha_s0_step100000_raw_g3.png", "agatha_v2_coco256_agatha_s0_step100000_ema_g1.png", "agatha_v2_coco256_agatha_s0_step100000_ema_g3.png", "agatha_v2_coco256_agatha_s0_evals.jsonl", "agatha_v2_coco256_agatha_s0_train.log"], "pushed": true}
|