Views
No views yet
| metric | value |
|---|---|
step | 190734 |
smooth_train_loss | 2.6008474826812744 |
min_objective | 0.7631417540553578 |
flops_used | 2.8688909511112733e+20 |
flops_per_token | 5737807872.0 |
total_training_time | 86673.26309394836 |
1{
2 "model_pt": {
3 "sequence_len": 2048,
4 "vocab_size": 65536,
5 "n_layer": 24,
6 "n_head": 12,
7 "n_kv_head": 12,
8 "n_embd": 1536
9 },
10 "model_ppt": null,
11 "optim": {
12 "matrix_lr": 0.006,
13 "embedding_lr": 0.3,
14 "unembedding_lr": 0.004,
15 "weight_decay": 0.0
16 },
17 "train": {
18 "num_iterations": 1000,
19 "eval_every_n_steps": 4000,
20 "eval_every_flops_frac": null,
21 "save_every_n_steps": null,
22 "save_at_steps": [
23 19073,
24 38147,
25 57220,
26 76294,
27 95367
28 ],
29 "log_every_n_steps": 100,
30 "eval_at_end": true,
31 "save_at_end": true,
32 "grad_clip": 1.0,
33 "ema_beta": 0.0
34 },
35 "eval": {
36 "eval_steps": 8,
37 "eval_tokens": 2016000
38 },
39 "hardware": {
40 "device_batch_size": 32,
41 "grad_accum_steps": 1,
42 "peak_tflops": 2250.0
43 },
44 "wandb": {
45 "enabled": true,
46 "notes": "",
47 "group": "5e10tok_vanilla",
48 "tags": [
49 "nanochat_gpt",
50 "pt_fineweb-nanochatbpe-55B",
51 "no_ppt",
52 "seed_0",
53 "single",
54 "lr_trapezoid",
55 "depth_24"
56 ],
57 "entity": null
58 },
59 "data_pt": "fineweb-nanochatbpe-55B",
60 "data_ppt": null,
61 "extra_eval": [
62 "c4-nanochatbpe-10B"
63 ],
64 "train_split": "train",
65 "pad_vocab": 65536,
66 "ppt_vocab_size": null,
67 "ppt_same_vocab_as_pt": false,
68 "ppt_unified_lr": false,
69 "lr_kind": "trapezoid",
70 "lr_warmup_ratio": 0.0,
71 "lr_warmdown_ratio": 0.5,
72 "lr_final_frac": 0.0,
73 "ppt_lr_kind": "trapezoid",
74 "ppt_lr_warmup_ratio": null,
75 "ppt_lr_warmdown_ratio": null,
76 "ppt_lr_final_frac": 0.0,
77 "ppt_lr": null,
78 "ppt_weight_decay": 0.0,
79 "ppt_device_batch_size": null,
80 "ppt_grad_accum_steps": null,
81 "eval_tokens": 2016000,
82 "reinit_embed_at_transition": true,
83 "moment_match_embed_reinit": false,
84 "reset_optimizer_at_transition": true,
85 "random_pt_head": false,
86 "depth": 24,
87 "compile_model": true,
88 "model_project": "longhorizon_50B_v0",
89 "run_name": "vanilla_50B_d24_seed0_resume38147",
90 "target_flops": 0.0,
91 "ppt_tokens": null,
92 "pt_tokens": 50000000000,
93 "ppt_eval_every_n_steps": null,
94 "alpha_ppt": 0.0,
95 "use_measured_flops": true,
96 "flops_per_token": null,
97 "ppt_flops_per_token": null,
98 "seed": 0,
99 "metrics_output_file": null,
100 "push_to_hf": true,
101 "hf_repo_org": "alexkstern",
102 "cleanup_checkpoints": false,
103 "resume_dir": "/root/longhorizon_checkpoints/wandb/longhorizon_50B_v0/vanilla_50B_d24_seed0_2026-07-25_11-16-20_588121/pt",
104 "resume_step": 38147
105}model_190734.pt — model weights (state_dict).meta_190734.json — training metadata.config_190734.json — run config snapshot.rng_190734.pt — RNG state (when present).