Upload logs/nanochat-tr-d20-bpe32k-492421.err
Browse files
logs/nanochat-tr-d20-bpe32k-492421.err
ADDED
|
@@ -0,0 +1,133 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Bu module sadece a100q ve a100x4q kuyrugundaki makinalarda calisir.
|
| 2 |
+
2026-06-07 16:23:31,528 - nanochat.common - [32m[1mINFO[0m - Distributed world size: 4
|
| 3 |
+
wandb: Currently logged in as: nurcunal (nurcunal-bogaziciuniversitesi) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
|
| 4 |
+
wandb: setting up run tr-d20-bpe32k-20260607-162301
|
| 5 |
+
wandb: Tracking run with wandb version 0.22.3
|
| 6 |
+
wandb: Run data is saved locally in /ari/users/nunal/nanochat-turk-d20-bpe32k/wandb/wandb/run-20260607_162332-tr-d20-bpe32k-20260607-162301
|
| 7 |
+
wandb: Run `wandb offline` to turn off syncing.
|
| 8 |
+
wandb: Syncing run tr-d20-bpe32k
|
| 9 |
+
wandb: βοΈ View project at https://wandb.ai/nurcunal-bogaziciuniversitesi/nanochat-turk
|
| 10 |
+
wandb: π View run at https://wandb.ai/nurcunal-bogaziciuniversitesi/nanochat-turk/runs/tr-d20-bpe32k-20260607-162301
|
| 11 |
+
2026-06-07 19:31:02,986 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_002500_rank2.pt
|
| 12 |
+
2026-06-07 19:31:03,721 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_002500_rank3.pt
|
| 13 |
+
2026-06-07 19:31:03,723 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_002500_rank1.pt
|
| 14 |
+
2026-06-07 19:31:04,669 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved model parameters to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/model_002500.pt
|
| 15 |
+
2026-06-07 19:31:04,672 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved metadata to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/meta_002500.json
|
| 16 |
+
2026-06-07 19:31:05,600 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_002500_rank0.pt
|
| 17 |
+
2026-06-07 22:36:52,330 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_005000_rank1.pt
|
| 18 |
+
2026-06-07 22:36:52,505 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_005000_rank2.pt
|
| 19 |
+
2026-06-07 22:36:52,614 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_005000_rank3.pt
|
| 20 |
+
2026-06-07 22:37:05,783 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved model parameters to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/model_005000.pt
|
| 21 |
+
2026-06-07 22:37:05,790 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved metadata to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/meta_005000.json
|
| 22 |
+
2026-06-07 22:37:07,259 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_005000_rank0.pt
|
| 23 |
+
2026-06-08 01:42:55,280 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_007500_rank2.pt
|
| 24 |
+
2026-06-08 01:42:55,666 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_007500_rank1.pt
|
| 25 |
+
2026-06-08 01:42:55,769 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_007500_rank3.pt
|
| 26 |
+
2026-06-08 01:42:58,578 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved model parameters to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/model_007500.pt
|
| 27 |
+
2026-06-08 01:42:58,582 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved metadata to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/meta_007500.json
|
| 28 |
+
2026-06-08 01:43:00,074 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_007500_rank0.pt
|
| 29 |
+
2026-06-08 04:48:53,363 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_010000_rank2.pt
|
| 30 |
+
2026-06-08 04:48:53,534 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_010000_rank1.pt
|
| 31 |
+
2026-06-08 04:48:56,328 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_010000_rank3.pt
|
| 32 |
+
2026-06-08 04:48:56,685 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved model parameters to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/model_010000.pt
|
| 33 |
+
2026-06-08 04:48:56,689 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved metadata to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/meta_010000.json
|
| 34 |
+
2026-06-08 04:49:01,264 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_010000_rank0.pt
|
| 35 |
+
2026-06-08 07:54:39,968 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_012500_rank3.pt
|
| 36 |
+
2026-06-08 07:54:39,991 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_012500_rank2.pt
|
| 37 |
+
2026-06-08 07:54:40,054 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_012500_rank1.pt
|
| 38 |
+
2026-06-08 07:54:41,707 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved model parameters to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/model_012500.pt
|
| 39 |
+
2026-06-08 07:54:41,710 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved metadata to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/meta_012500.json
|
| 40 |
+
2026-06-08 07:54:42,841 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_012500_rank0.pt
|
| 41 |
+
2026-06-08 11:00:33,210 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_015000_rank1.pt
|
| 42 |
+
2026-06-08 11:00:33,211 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_015000_rank2.pt
|
| 43 |
+
2026-06-08 11:00:33,637 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_015000_rank3.pt
|
| 44 |
+
2026-06-08 11:00:47,652 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved model parameters to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/model_015000.pt
|
| 45 |
+
2026-06-08 11:00:47,655 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved metadata to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/meta_015000.json
|
| 46 |
+
2026-06-08 11:00:53,283 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_015000_rank0.pt
|
| 47 |
+
2026-06-08 13:37:24,481 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_017100_rank2.pt
|
| 48 |
+
2026-06-08 13:37:24,556 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_017100_rank1.pt
|
| 49 |
+
2026-06-08 13:37:24,815 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_017100_rank3.pt
|
| 50 |
+
2026-06-08 13:37:28,031 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved model parameters to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/model_017100.pt
|
| 51 |
+
2026-06-08 13:37:28,038 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved metadata to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/meta_017100.json
|
| 52 |
+
2026-06-08 13:37:29,349 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Saved optimizer state to: /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20/optim_017100_rank0.pt
|
| 53 |
+
wandb: updating run metadata
|
| 54 |
+
wandb: uploading output.log; uploading wandb-summary.json
|
| 55 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 56 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading history steps 189-189, summary, console lines 17323-17336
|
| 57 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 58 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 59 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 60 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 61 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 62 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 63 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 64 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 65 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 66 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 67 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 68 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 69 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 70 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 71 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 72 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 73 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 74 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 75 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 76 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 77 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 78 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 79 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 80 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 81 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 82 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 83 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 84 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 85 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 86 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 87 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 88 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 89 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 90 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 91 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 92 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 93 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 94 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 95 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 96 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 97 |
+
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
|
| 98 |
+
wandb: uploading wandb-summary.json; uploading config.yaml
|
| 99 |
+
wandb: uploading wandb-summary.json; uploading config.yaml; uploading data
|
| 100 |
+
wandb: uploading wandb-summary.json; uploading config.yaml
|
| 101 |
+
wandb: uploading config.yaml
|
| 102 |
+
wandb: uploading data
|
| 103 |
+
wandb:
|
| 104 |
+
wandb: Run history:
|
| 105 |
+
wandb: step βββββββββββββββββββββ
β
β
β
β
β
β
β
ββββββββββββ
|
| 106 |
+
wandb: total_training_flops βββββββββββββββββ
β
β
β
β
β
ββββββββββββββββββ
|
| 107 |
+
wandb: total_training_time ββββββββββββββββββββββββ
β
β
β
βββββββββββββ
|
| 108 |
+
wandb: train/dt β
βββ
βββββββββββββ
βββββββββββββββββββββββ
|
| 109 |
+
wandb: train/loss ββββββββββββββββββββββββββββββββββββββββ
|
| 110 |
+
wandb: train/lrm βββββββββββββββββββββ
β
β
βββββββββββββββββ
|
| 111 |
+
wandb: train/mfu ββββββββββββββββββββββββββββββββββββββββ
|
| 112 |
+
wandb: train/tok_per_sec ββββββββββββββββββββββββββββββββββββββββ
|
| 113 |
+
wandb: val/bpb βββββββββββββββββββ
|
| 114 |
+
wandb:
|
| 115 |
+
wandb: Run summary:
|
| 116 |
+
wandb: step 17100
|
| 117 |
+
wandb: total_training_flops 5.809722400073318e+19
|
| 118 |
+
wandb: total_training_time 76059.47399
|
| 119 |
+
wandb: train/dt 4.43947
|
| 120 |
+
wandb: train/epoch 1 pq: 10 rg: 1744
|
| 121 |
+
wandb: train/loss 2.38038
|
| 122 |
+
wandb: train/lrm 0.05855
|
| 123 |
+
wandb: train/mfu 61.32155
|
| 124 |
+
wandb: train/tok_per_sec 236193
|
| 125 |
+
wandb: val/bpb 0.62321
|
| 126 |
+
wandb:
|
| 127 |
+
wandb: π View run tr-d20-bpe32k at: https://wandb.ai/nurcunal-bogaziciuniversitesi/nanochat-turk/runs/tr-d20-bpe32k-20260607-162301
|
| 128 |
+
wandb: βοΈ View project at: https://wandb.ai/nurcunal-bogaziciuniversitesi/nanochat-turk
|
| 129 |
+
wandb: Synced 5 W&B file(s), 0 media file(s), 0 artifact file(s) and 0 other file(s)
|
| 130 |
+
wandb: Find logs at: .-d20-bpe32k/wandb/wandb/run-20260607_162332-tr-d20-bpe32k-20260607-162301/logs
|
| 131 |
+
2026-06-08 13:54:06,931 - nanochat.common - [32m[1mINFO[0m - Distributed world size: 4
|
| 132 |
+
2026-06-08 13:54:06,934 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Loading model from /ari/users/nunal/nanochat-turk-d20-bpe32k/base_checkpoints/tr_d20_bpe_32768_chinchilla20 with step 17100
|
| 133 |
+
2026-06-08 13:54:08,704 - nanochat.checkpoint_manager - [32m[1mINFO[0m - Building model with config: {'sequence_len': 2048, 'vocab_size': 32768, 'n_layer': 20, 'n_head': 10, 'n_kv_head': 10, 'n_embd': 1280, 'window_pattern': 'L'}
|