Hugging Face's logo

Skywalker1910
/
BB8

Text Generation
Transformers
Safetensors
PyTorch
English
language-model
from-scratch
lora
educational
gpt
decoder-only
Model card Files
xet
 Community
Skywalker1910 commited on
Commit
8e1700a
·
verified ·
1 Parent(s): 6690f40

Upload folder using huggingface_hub

Browse files
Files changed (2) hide show
  1. outputs/bb8-char-small-v002/results.json +109 -0
  2. outputs/bb8-char-small-v002/run.json +69 -0
outputs/bb8-char-small-v002/results.json ADDED
@@ -0,0 +1,109 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "experiment_name": "bb8-char-small-v002",
3
+ "config": {
4
+ "model": {
5
+ "d_model": 64,
6
+ "num_layers": 2,
7
+ "num_heads": 2,
8
+ "d_ff": 256,
9
+ "max_seq_len": 128,
10
+ "dropout": 0.1,
11
+ "activation": "gelu",
12
+ "tie_weights": true
13
+ },
14
+ "tokenizer": {
15
+ "type": "char",
16
+ "vocab_size": null
17
+ },
18
+ "training": {
19
+ "batch_size": 64,
20
+ "num_epochs": 5,
21
+ "lr": 0.001,
22
+ "weight_decay": 0.1,
23
+ "max_grad_norm": 1.0,
24
+ "warmup_steps": 50,
25
+ "mixed_precision": false,
26
+ "log_interval": 100,
27
+ "eval_interval": 0,
28
+ "save_interval": 5000
29
+ },
30
+ "data": {
31
+ "seq_len": 128,
32
+ "stride": 1,
33
+ "val_fraction": 0.1
34
+ },
35
+ "generation": {
36
+ "max_new_tokens": 100,
37
+ "temperature": 0.8,
38
+ "top_k": 40,
39
+ "top_p": 0.9,
40
+ "repetition_penalty": 1.1
41
+ }
42
+ },
43
+ "tokenizer": {
44
+ "type": "char",
45
+ "vocab_size": 69
46
+ },
47
+ "model_params": 112320,
48
+ "model_config": {
49
+ "vocab_size": 69,
50
+ "d_model": 64,
51
+ "num_layers": 2,
52
+ "num_heads": 2,
53
+ "d_ff": 256,
54
+ "max_seq_len": 128,
55
+ "dropout": 0.1,
56
+ "activation": "gelu",
57
+ "tie_weights": true
58
+ },
59
+ "training_history": {
60
+ "train_loss": [
61
+ 1.9597480971617243,
62
+ 1.760318691349072,
63
+ 1.723512035695325,
64
+ 1.6932429660379844,
65
+ 1.6759626470581728
66
+ ],
67
+ "val_loss": [
68
+ 1.7861656403144428,
69
+ 1.7393092440846731,
70
+ 1.7209157428133426,
71
+ 1.694505056179097,
72
+ 1.6863457313698642
73
+ ],
74
+ "perplexity": [
75
+ 5.9665307244925145,
76
+ 5.69340930791408,
77
+ 5.589644798348661,
78
+ 5.443950848468378,
79
+ 5.399712607765923
80
+ ],
81
+ "lr": [
82
+ 0.0009144813286784496,
83
+ 0.0006895722629658809,
84
+ 0.00041128554131832963,
85
+ 0.0001860484191497777,
86
+ 0.0001
87
+ ]
88
+ },
89
+ "evaluation": {
90
+ "train": {
91
+ "loss": 1.5084002385491857,
92
+ "perplexity": 4.5194948942914515,
93
+ "accuracy": 0.5404959213275322,
94
+ "bits_per_char": 2.1761615438306396
95
+ },
96
+ "val": {
97
+ "loss": 1.6863457313698642,
98
+ "perplexity": 5.399712607765923,
99
+ "accuracy": 0.4991981479777762,
100
+ "bits_per_char": 2.432882623871575
101
+ }
102
+ },
103
+ "generated_samples": {
104
+ "greedy": "ROMEO:\nThe shall the come the shall be the shall the prince the shall the shall the shall the some the sha",
105
+ "temperature": "ROMEO:\nHo die.\n\nKING RICHARD III:\nI go that and my brothers, it parts fed!\n\nPRINCE:\nWho, but where's that ",
106
+ "top_k": "ROMEO:\nMy do live not breads a maving.\n\nDUKE VINCENTIO:\nGrace, not within the suffed that this riffence\nAn",
107
+ "top_p": "ROMEO:\nThere but done.\n\nROMEO:\nWhat, shall proming, and they have to be better,\nAnd make that and to heart"
108
+ }
109
+ }
outputs/bb8-char-small-v002/run.json ADDED
@@ -0,0 +1,69 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "run_id": "bb8-char-small-v002",
3
+ "started_at": "2026-09-25T01:50:29.960997+00:00",
4
+ "completed_at": "2026-09-25T02:11:54.360979+00:00",
5
+ "duration_seconds": 1284.4,
6
+ "git_commit": "5e22799e0b14096994bfe5b2f4432344ab31911b",
7
+ "git_dirty": true,
8
+ "dataset": {
9
+ "path": "data\\tiny_shakespeare.txt",
10
+ "sha256": "86c4e6aa9db7c042ec79f339dcb96d42b0075e16b8fc2e86bf0ca57e2dc565ed",
11
+ "characters": 1115394
12
+ },
13
+ "tokenizer": {
14
+ "type": "char",
15
+ "vocab_size": 69
16
+ },
17
+ "model": {
18
+ "parameters": 112320,
19
+ "vocab_size": 69,
20
+ "d_model": 64,
21
+ "num_layers": 2,
22
+ "num_heads": 2,
23
+ "d_ff": 256,
24
+ "max_seq_len": 128,
25
+ "dropout": 0.1,
26
+ "activation": "gelu",
27
+ "tie_weights": true
28
+ },
29
+ "training": {
30
+ "batch_size": 64,
31
+ "num_epochs": 5,
32
+ "lr": 0.001,
33
+ "weight_decay": 0.1,
34
+ "max_grad_norm": 1.0,
35
+ "warmup_steps": 50,
36
+ "mixed_precision": false,
37
+ "log_interval": 100,
38
+ "eval_interval": 0,
39
+ "save_interval": 5000
40
+ },
41
+ "runtime": {
42
+ "device": "cuda",
43
+ "pytorch": "2.13.0+cu130",
44
+ "cuda": "13.0",
45
+ "gpu": "NVIDIA GeForce RTX 4070 Laptop GPU",
46
+ "gpu_memory_gb": 8.0
47
+ },
48
+ "evaluation": {
49
+ "train": {
50
+ "loss": 1.5084002385491857,
51
+ "perplexity": 4.5194948942914515,
52
+ "accuracy": 0.5404959213275322,
53
+ "bits_per_char": 2.1761615438306396
54
+ },
55
+ "val": {
56
+ "loss": 1.6863457313698642,
57
+ "perplexity": 5.399712607765923,
58
+ "accuracy": 0.4991981479777762,
59
+ "bits_per_char": 2.432882623871575
60
+ }
61
+ },
62
+ "artifacts": {
63
+ "checkpoint": "checkpoints/bb8-char-small-v002/final_model.pt",
64
+ "tokenizer": "checkpoints/bb8-char-small-v002/tokenizer.json",
65
+ "config": "checkpoints/bb8-char-small-v002/config.yaml",
66
+ "results": "outputs/bb8-char-small-v002/results.json",
67
+ "output_dir": "outputs\\bb8-char-small-v002"
68
+ }
69
+ }