plip commited on
Commit
0145f9c
1 Parent(s): b14e21e

Training in progress, step 10000

Browse files
.gitattributes CHANGED
@@ -32,3 +32,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
32
  *.zip filter=lfs diff=lfs merge=lfs -text
33
  *.zst filter=lfs diff=lfs merge=lfs -text
34
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
32
  *.zip filter=lfs diff=lfs merge=lfs -text
33
  *.zst filter=lfs diff=lfs merge=lfs -text
34
  *tfevents* filter=lfs diff=lfs merge=lfs -text
35
+ NotoSansMono-Regular.ttf filter=lfs diff=lfs merge=lfs -text
.gitignore ADDED
@@ -0,0 +1 @@
 
 
1
+ checkpoint-*/
NotoSansMono-Regular.ttf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:97b31969bfe7ebf894c12afdf6a5dcb6e87e64114697782802fe933e693eacd8
3
+ size 383844
config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "PIXELForPreTraining"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.1,
6
+ "decoder_hidden_size": 264,
7
+ "decoder_intermediate_size": 1056,
8
+ "decoder_num_attention_heads": 6,
9
+ "decoder_num_hidden_layers": 4,
10
+ "hidden_act": "gelu",
11
+ "hidden_dropout_prob": 0.1,
12
+ "hidden_size": 384,
13
+ "image_size": [
14
+ 16,
15
+ 8464
16
+ ],
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 1536,
19
+ "layer_norm_eps": 1e-12,
20
+ "mask_ratio": 0.25,
21
+ "model_type": "pixel",
22
+ "norm_pix_loss": true,
23
+ "num_attention_heads": 6,
24
+ "num_channels": 3,
25
+ "num_hidden_layers": 12,
26
+ "patch_size": 16,
27
+ "qkv_bias": true,
28
+ "torch_dtype": "float32",
29
+ "transformers_version": "4.17.0"
30
+ }
last-checkpoint/NotoSansMono-Regular.ttf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:97b31969bfe7ebf894c12afdf6a5dcb6e87e64114697782802fe933e693eacd8
3
+ size 383844
last-checkpoint/config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "PIXELForPreTraining"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.1,
6
+ "decoder_hidden_size": 264,
7
+ "decoder_intermediate_size": 1056,
8
+ "decoder_num_attention_heads": 6,
9
+ "decoder_num_hidden_layers": 4,
10
+ "hidden_act": "gelu",
11
+ "hidden_dropout_prob": 0.1,
12
+ "hidden_size": 384,
13
+ "image_size": [
14
+ 16,
15
+ 8464
16
+ ],
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 1536,
19
+ "layer_norm_eps": 1e-12,
20
+ "mask_ratio": 0.25,
21
+ "model_type": "pixel",
22
+ "norm_pix_loss": true,
23
+ "num_attention_heads": 6,
24
+ "num_channels": 3,
25
+ "num_hidden_layers": 12,
26
+ "patch_size": 16,
27
+ "qkv_bias": true,
28
+ "torch_dtype": "float32",
29
+ "transformers_version": "4.17.0"
30
+ }
last-checkpoint/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7a9b46ed6af836e8924481b9ef3b8ac908f671a8de3334b85575e607c66e5a72
3
+ size 202193937
last-checkpoint/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9fa1850686699db6f7d38adef4e1f7555c082f72c52d9543757d96d6e7dfe575
3
+ size 102501541
last-checkpoint/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f63b6e7591877af348d0715da8e499a40da21a9ac10ea61d3d17f2882b7374e
3
+ size 14503
last-checkpoint/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f63b6e7591877af348d0715da8e499a40da21a9ac10ea61d3d17f2882b7374e
3
+ size 14503
last-checkpoint/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f63b6e7591877af348d0715da8e499a40da21a9ac10ea61d3d17f2882b7374e
3
+ size 14503
last-checkpoint/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f63b6e7591877af348d0715da8e499a40da21a9ac10ea61d3d17f2882b7374e
3
+ size 14503
last-checkpoint/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f63b6e7591877af348d0715da8e499a40da21a9ac10ea61d3d17f2882b7374e
3
+ size 14503
last-checkpoint/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f63b6e7591877af348d0715da8e499a40da21a9ac10ea61d3d17f2882b7374e
3
+ size 14503
last-checkpoint/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f63b6e7591877af348d0715da8e499a40da21a9ac10ea61d3d17f2882b7374e
3
+ size 14503
last-checkpoint/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f63b6e7591877af348d0715da8e499a40da21a9ac10ea61d3d17f2882b7374e
3
+ size 14503
last-checkpoint/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4931c97be633a6e3808391ac03b792837805ded64ff2a578babfa79e60dc22a7
3
+ size 623
last-checkpoint/text_renderer_config.json ADDED
@@ -0,0 +1,224 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "background_color": "white",
3
+ "disable_fallback": false,
4
+ "dpi": 120,
5
+ "font_color": "black",
6
+ "font_file": "NotoSansMono-Regular.ttf",
7
+ "font_size": 8,
8
+ "fonts_list": [
9
+ "Apple Color Emoji",
10
+ "C059",
11
+ "D050000L",
12
+ "DejaVu Sans",
13
+ "Inconsolata",
14
+ "Monospace",
15
+ "Nimbus Mono PS",
16
+ "Nimbus Roman",
17
+ "Nimbus Sans",
18
+ "Nimbus Sans Narrow",
19
+ "Noto Kufi Arabic",
20
+ "Noto Looped Lao",
21
+ "Noto Looped Lao UI",
22
+ "Noto Music",
23
+ "Noto Naskh Arabic",
24
+ "Noto Naskh Arabic UI",
25
+ "Noto Nastaliq Urdu",
26
+ "Noto Rashi Hebrew",
27
+ "Noto Sans Adlam",
28
+ "Noto Sans Adlam Unjoined",
29
+ "Noto Sans Anatolian Hieroglyphs",
30
+ "Noto Sans Arabic",
31
+ "Noto Sans Armenian",
32
+ "Noto Sans Avestan",
33
+ "Noto Sans Balinese",
34
+ "Noto Sans Bamum",
35
+ "Noto Sans Bassa Vah",
36
+ "Noto Sans Batak",
37
+ "Noto Sans Bengali",
38
+ "Noto Sans Bengali UI",
39
+ "Noto Sans Bhaiksuki",
40
+ "Noto Sans Brahmi",
41
+ "Noto Sans Buginese",
42
+ "Noto Sans Buhid",
43
+ "Noto Sans CJK HK",
44
+ "Noto Sans CJK JP",
45
+ "Noto Sans CJK KR",
46
+ "Noto Sans CJK SC",
47
+ "Noto Sans CJK TC",
48
+ "Noto Sans Canadian Aboriginal",
49
+ "Noto Sans Carian",
50
+ "Noto Sans Caucasian Albanian",
51
+ "Noto Sans Chakma",
52
+ "Noto Sans Cham",
53
+ "Noto Sans Cherokee",
54
+ "Noto Sans Chorasmian",
55
+ "Noto Sans Coptic",
56
+ "Noto Sans Cuneiform",
57
+ "Noto Sans Cypriot",
58
+ "Noto Sans Cypro Minoan",
59
+ "Noto Sans Deseret",
60
+ "Noto Sans Devanagari",
61
+ "Noto Sans Devanagari UI",
62
+ "Noto Sans Duployan",
63
+ "Noto Sans Egyptian Hieroglyphs",
64
+ "Noto Sans Elbasan",
65
+ "Noto Sans Elymaic",
66
+ "Noto Sans Ethiopic",
67
+ "Noto Sans Georgian",
68
+ "Noto Sans Glagolitic",
69
+ "Noto Sans Gothic",
70
+ "Noto Sans Grantha",
71
+ "Noto Sans Gujarati",
72
+ "Noto Sans Gujarati UI",
73
+ "Noto Sans Gunjala Gondi",
74
+ "Noto Sans Gurmukhi",
75
+ "Noto Sans Gurmukhi UI",
76
+ "Noto Sans Hanifi Rohingya",
77
+ "Noto Sans Hanunoo",
78
+ "Noto Sans Hatran",
79
+ "Noto Sans Hebrew",
80
+ "Noto Sans Hebrew Droid",
81
+ "Noto Sans Imperial Aramaic",
82
+ "Noto Sans Indic Siyaq Numbers",
83
+ "Noto Sans Inscriptional Pahlavi",
84
+ "Noto Sans Inscriptional Parthian",
85
+ "Noto Sans Javanese",
86
+ "Noto Sans Kaithi",
87
+ "Noto Sans Kannada",
88
+ "Noto Sans Kannada UI",
89
+ "Noto Sans Kayah Li",
90
+ "Noto Sans Kharoshthi",
91
+ "Noto Sans Khmer",
92
+ "Noto Sans Khmer UI",
93
+ "Noto Sans Khudawadi",
94
+ "Noto Sans Lao",
95
+ "Noto Sans Lao UI",
96
+ "Noto Sans Lepcha",
97
+ "Noto Sans Limbu",
98
+ "Noto Sans Linear A",
99
+ "Noto Sans Linear B",
100
+ "Noto Sans Lisu",
101
+ "Noto Sans Lycian",
102
+ "Noto Sans Lydian",
103
+ "Noto Sans Mahajani",
104
+ "Noto Sans Malayalam",
105
+ "Noto Sans Malayalam UI",
106
+ "Noto Sans Mandaic",
107
+ "Noto Sans Manichaean",
108
+ "Noto Sans Marchen",
109
+ "Noto Sans Masaram Gondi",
110
+ "Noto Sans Math",
111
+ "Noto Sans Mayan Numerals",
112
+ "Noto Sans Medefaidrin",
113
+ "Noto Sans Meetei Mayek",
114
+ "Noto Sans Mende Kikakui",
115
+ "Noto Sans Meroitic",
116
+ "Noto Sans Miao",
117
+ "Noto Sans Modi",
118
+ "Noto Sans Mongolian",
119
+ "Noto Sans Mono",
120
+ "Noto Sans Mro",
121
+ "Noto Sans Multani",
122
+ "Noto Sans Myanmar",
123
+ "Noto Sans Myanmar UI",
124
+ "Noto Sans NKo",
125
+ "Noto Sans Nabataean",
126
+ "Noto Sans Nandinagari",
127
+ "Noto Sans New Tai Lue",
128
+ "Noto Sans Newa",
129
+ "Noto Sans Nushu",
130
+ "Noto Sans Ogham",
131
+ "Noto Sans Old",
132
+ "Noto Sans Old Hungarian",
133
+ "Noto Sans Old North Arabian",
134
+ "Noto Sans Old Permic",
135
+ "Noto Sans Old Persian",
136
+ "Noto Sans Old Sogdian",
137
+ "Noto Sans Old South Arabian",
138
+ "Noto Sans Old Turkic",
139
+ "Noto Sans Oriya",
140
+ "Noto Sans Osage",
141
+ "Noto Sans Osmanya",
142
+ "Noto Sans Pahawh Hmong",
143
+ "Noto Sans Palmyrene",
144
+ "Noto Sans Pau Cin Hau",
145
+ "Noto Sans PhagsPa",
146
+ "Noto Sans Phoenician",
147
+ "Noto Sans Psalter Pahlavi",
148
+ "Noto Sans Rejang",
149
+ "Noto Sans Runic",
150
+ "Noto Sans Samaritan",
151
+ "Noto Sans Saurashtra",
152
+ "Noto Sans Sharada",
153
+ "Noto Sans Shavian",
154
+ "Noto Sans Siddham",
155
+ "Noto Sans SignWriting",
156
+ "Noto Sans Sinhala",
157
+ "Noto Sans Sinhala UI",
158
+ "Noto Sans Sogdian",
159
+ "Noto Sans Soyombo",
160
+ "Noto Sans Syloti Nagri",
161
+ "Noto Sans Symbols",
162
+ "Noto Sans Symbols2",
163
+ "Noto Sans Syriac",
164
+ "Noto Sans Tagalog",
165
+ "Noto Sans Tagbanwa",
166
+ "Noto Sans Tai Le",
167
+ "Noto Sans Tai Tham",
168
+ "Noto Sans Tai Viet",
169
+ "Noto Sans Takri",
170
+ "Noto Sans Tamil",
171
+ "Noto Sans Tamil Supplement",
172
+ "Noto Sans Tamil UI",
173
+ "Noto Sans Tangsa",
174
+ "Noto Sans Telugu",
175
+ "Noto Sans Telugu UI",
176
+ "Noto Sans Test",
177
+ "Noto Sans Thaana",
178
+ "Noto Sans Thai",
179
+ "Noto Sans Thai Looped",
180
+ "Noto Sans Thai Looped UI",
181
+ "Noto Sans Thai UI",
182
+ "Noto Sans Tifinagh",
183
+ "Noto Sans Tifinagh APT",
184
+ "Noto Sans Tifinagh Adrar",
185
+ "Noto Sans Tifinagh Agraw Imazighen",
186
+ "Noto Sans Tifinagh Ahaggar",
187
+ "Noto Sans Tifinagh Air",
188
+ "Noto Sans Tifinagh Azawagh",
189
+ "Noto Sans Tifinagh Ghat",
190
+ "Noto Sans Tifinagh Hawad",
191
+ "Noto Sans Tifinagh Rhissa Ixa",
192
+ "Noto Sans Tifinagh SIL",
193
+ "Noto Sans Tifinagh Tawellemmet",
194
+ "Noto Sans Tirhuta",
195
+ "Noto Sans Ugaritic",
196
+ "Noto Sans Vai",
197
+ "Noto Sans Vithkuqi",
198
+ "Noto Sans Wancho",
199
+ "Noto Sans Warang Citi",
200
+ "Noto Sans Yi",
201
+ "Noto Sans Zanabazar Square",
202
+ "Noto Traditional Nushu",
203
+ "P052",
204
+ "Sans",
205
+ "Serif",
206
+ "Source Code Pro",
207
+ "System-ui",
208
+ "URW Bookman",
209
+ "URW Gothic",
210
+ "Ubuntu",
211
+ "Ubuntu Condensed",
212
+ "Ubuntu Mono",
213
+ "Z003",
214
+ "Go Noto Current"
215
+ ],
216
+ "max_seq_length": 529,
217
+ "non_monospace_dpi": 120,
218
+ "non_monospace_font_file": "/scratch/project/dd-22-48/code/monospace_experiments_scripts/monospace_renderer/GoNotoCurrent.ttf",
219
+ "non_monospace_font_size": 8,
220
+ "pad_size": 3,
221
+ "pixels_per_patch": 16,
222
+ "rgb": true,
223
+ "text_renderer_type": "MonoPangoCairoTextRenderer"
224
+ }
last-checkpoint/trainer_state.json ADDED
@@ -0,0 +1,216 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.23299704093758009,
5
+ "global_step": 10000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.01,
12
+ "learning_rate": 5.999999999999999e-06,
13
+ "loss": 0.9072,
14
+ "step": 500
15
+ },
16
+ {
17
+ "epoch": 0.02,
18
+ "learning_rate": 1.1999999999999999e-05,
19
+ "loss": 0.6779,
20
+ "step": 1000
21
+ },
22
+ {
23
+ "epoch": 0.02,
24
+ "eval_loss": 0.6512336730957031,
25
+ "eval_runtime": 1.3899,
26
+ "eval_samples_per_second": 1581.436,
27
+ "eval_steps_per_second": 25.182,
28
+ "step": 1000
29
+ },
30
+ {
31
+ "epoch": 0.03,
32
+ "learning_rate": 1.7999999999999997e-05,
33
+ "loss": 0.6516,
34
+ "step": 1500
35
+ },
36
+ {
37
+ "epoch": 0.05,
38
+ "learning_rate": 2.3999999999999997e-05,
39
+ "loss": 0.6505,
40
+ "step": 2000
41
+ },
42
+ {
43
+ "epoch": 0.05,
44
+ "eval_loss": 0.6479398608207703,
45
+ "eval_runtime": 1.2819,
46
+ "eval_samples_per_second": 1714.632,
47
+ "eval_steps_per_second": 27.303,
48
+ "step": 2000
49
+ },
50
+ {
51
+ "epoch": 0.06,
52
+ "learning_rate": 2.9999999999999997e-05,
53
+ "loss": 0.6487,
54
+ "step": 2500
55
+ },
56
+ {
57
+ "epoch": 0.07,
58
+ "learning_rate": 3.5999999999999994e-05,
59
+ "loss": 0.6475,
60
+ "step": 3000
61
+ },
62
+ {
63
+ "epoch": 0.07,
64
+ "eval_loss": 0.6461147665977478,
65
+ "eval_runtime": 1.2932,
66
+ "eval_samples_per_second": 1699.633,
67
+ "eval_steps_per_second": 27.064,
68
+ "step": 3000
69
+ },
70
+ {
71
+ "epoch": 0.08,
72
+ "learning_rate": 4.2e-05,
73
+ "loss": 0.6469,
74
+ "step": 3500
75
+ },
76
+ {
77
+ "epoch": 0.09,
78
+ "learning_rate": 4.7999999999999994e-05,
79
+ "loss": 0.6464,
80
+ "step": 4000
81
+ },
82
+ {
83
+ "epoch": 0.09,
84
+ "eval_loss": 0.6451290249824524,
85
+ "eval_runtime": 1.2355,
86
+ "eval_samples_per_second": 1779.085,
87
+ "eval_steps_per_second": 28.329,
88
+ "step": 4000
89
+ },
90
+ {
91
+ "epoch": 0.1,
92
+ "learning_rate": 5.399999999999999e-05,
93
+ "loss": 0.6459,
94
+ "step": 4500
95
+ },
96
+ {
97
+ "epoch": 0.12,
98
+ "learning_rate": 5.9999999999999995e-05,
99
+ "loss": 0.6455,
100
+ "step": 5000
101
+ },
102
+ {
103
+ "epoch": 0.12,
104
+ "eval_loss": 0.6444417238235474,
105
+ "eval_runtime": 1.2972,
106
+ "eval_samples_per_second": 1694.435,
107
+ "eval_steps_per_second": 26.981,
108
+ "step": 5000
109
+ },
110
+ {
111
+ "epoch": 0.13,
112
+ "learning_rate": 6.599999999999999e-05,
113
+ "loss": 0.6452,
114
+ "step": 5500
115
+ },
116
+ {
117
+ "epoch": 0.14,
118
+ "learning_rate": 7.199999999999999e-05,
119
+ "loss": 0.6448,
120
+ "step": 6000
121
+ },
122
+ {
123
+ "epoch": 0.14,
124
+ "eval_loss": 0.6438854336738586,
125
+ "eval_runtime": 1.2871,
126
+ "eval_samples_per_second": 1707.767,
127
+ "eval_steps_per_second": 27.194,
128
+ "step": 6000
129
+ },
130
+ {
131
+ "epoch": 0.15,
132
+ "learning_rate": 7.8e-05,
133
+ "loss": 0.64,
134
+ "step": 6500
135
+ },
136
+ {
137
+ "epoch": 0.16,
138
+ "learning_rate": 8.4e-05,
139
+ "loss": 0.6222,
140
+ "step": 7000
141
+ },
142
+ {
143
+ "epoch": 0.16,
144
+ "eval_loss": 0.6123253703117371,
145
+ "eval_runtime": 1.2501,
146
+ "eval_samples_per_second": 1758.21,
147
+ "eval_steps_per_second": 27.997,
148
+ "step": 7000
149
+ },
150
+ {
151
+ "epoch": 0.17,
152
+ "learning_rate": 8.999999999999999e-05,
153
+ "loss": 0.6115,
154
+ "step": 7500
155
+ },
156
+ {
157
+ "epoch": 0.19,
158
+ "learning_rate": 9.599999999999999e-05,
159
+ "loss": 0.6019,
160
+ "step": 8000
161
+ },
162
+ {
163
+ "epoch": 0.19,
164
+ "eval_loss": 0.5872684717178345,
165
+ "eval_runtime": 1.2778,
166
+ "eval_samples_per_second": 1720.181,
167
+ "eval_steps_per_second": 27.391,
168
+ "step": 8000
169
+ },
170
+ {
171
+ "epoch": 0.2,
172
+ "learning_rate": 0.000102,
173
+ "loss": 0.5924,
174
+ "step": 8500
175
+ },
176
+ {
177
+ "epoch": 0.21,
178
+ "learning_rate": 0.00010799999999999998,
179
+ "loss": 0.5838,
180
+ "step": 9000
181
+ },
182
+ {
183
+ "epoch": 0.21,
184
+ "eval_loss": 0.5689075589179993,
185
+ "eval_runtime": 1.2743,
186
+ "eval_samples_per_second": 1724.838,
187
+ "eval_steps_per_second": 27.466,
188
+ "step": 9000
189
+ },
190
+ {
191
+ "epoch": 0.22,
192
+ "learning_rate": 0.00011399999999999999,
193
+ "loss": 0.5737,
194
+ "step": 9500
195
+ },
196
+ {
197
+ "epoch": 0.23,
198
+ "learning_rate": 0.00011999999999999999,
199
+ "loss": 0.5612,
200
+ "step": 10000
201
+ },
202
+ {
203
+ "epoch": 0.23,
204
+ "eval_loss": 0.5473007559776306,
205
+ "eval_runtime": 1.2648,
206
+ "eval_samples_per_second": 1737.882,
207
+ "eval_steps_per_second": 27.673,
208
+ "step": 10000
209
+ }
210
+ ],
211
+ "max_steps": 500000,
212
+ "num_train_epochs": 12,
213
+ "total_flos": 3.194871387745e+20,
214
+ "trial_name": null,
215
+ "trial_params": null
216
+ }
last-checkpoint/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9836cbf6995d22ed625995c395e0a3b185cdcffe7981eeaf2436d50a2a13d201
3
+ size 3375
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9fa1850686699db6f7d38adef4e1f7555c082f72c52d9543757d96d6e7dfe575
3
+ size 102501541
text_renderer_config.json ADDED
@@ -0,0 +1,224 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "background_color": "white",
3
+ "disable_fallback": false,
4
+ "dpi": 120,
5
+ "font_color": "black",
6
+ "font_file": "NotoSansMono-Regular.ttf",
7
+ "font_size": 8,
8
+ "fonts_list": [
9
+ "Apple Color Emoji",
10
+ "C059",
11
+ "D050000L",
12
+ "DejaVu Sans",
13
+ "Inconsolata",
14
+ "Monospace",
15
+ "Nimbus Mono PS",
16
+ "Nimbus Roman",
17
+ "Nimbus Sans",
18
+ "Nimbus Sans Narrow",
19
+ "Noto Kufi Arabic",
20
+ "Noto Looped Lao",
21
+ "Noto Looped Lao UI",
22
+ "Noto Music",
23
+ "Noto Naskh Arabic",
24
+ "Noto Naskh Arabic UI",
25
+ "Noto Nastaliq Urdu",
26
+ "Noto Rashi Hebrew",
27
+ "Noto Sans Adlam",
28
+ "Noto Sans Adlam Unjoined",
29
+ "Noto Sans Anatolian Hieroglyphs",
30
+ "Noto Sans Arabic",
31
+ "Noto Sans Armenian",
32
+ "Noto Sans Avestan",
33
+ "Noto Sans Balinese",
34
+ "Noto Sans Bamum",
35
+ "Noto Sans Bassa Vah",
36
+ "Noto Sans Batak",
37
+ "Noto Sans Bengali",
38
+ "Noto Sans Bengali UI",
39
+ "Noto Sans Bhaiksuki",
40
+ "Noto Sans Brahmi",
41
+ "Noto Sans Buginese",
42
+ "Noto Sans Buhid",
43
+ "Noto Sans CJK HK",
44
+ "Noto Sans CJK JP",
45
+ "Noto Sans CJK KR",
46
+ "Noto Sans CJK SC",
47
+ "Noto Sans CJK TC",
48
+ "Noto Sans Canadian Aboriginal",
49
+ "Noto Sans Carian",
50
+ "Noto Sans Caucasian Albanian",
51
+ "Noto Sans Chakma",
52
+ "Noto Sans Cham",
53
+ "Noto Sans Cherokee",
54
+ "Noto Sans Chorasmian",
55
+ "Noto Sans Coptic",
56
+ "Noto Sans Cuneiform",
57
+ "Noto Sans Cypriot",
58
+ "Noto Sans Cypro Minoan",
59
+ "Noto Sans Deseret",
60
+ "Noto Sans Devanagari",
61
+ "Noto Sans Devanagari UI",
62
+ "Noto Sans Duployan",
63
+ "Noto Sans Egyptian Hieroglyphs",
64
+ "Noto Sans Elbasan",
65
+ "Noto Sans Elymaic",
66
+ "Noto Sans Ethiopic",
67
+ "Noto Sans Georgian",
68
+ "Noto Sans Glagolitic",
69
+ "Noto Sans Gothic",
70
+ "Noto Sans Grantha",
71
+ "Noto Sans Gujarati",
72
+ "Noto Sans Gujarati UI",
73
+ "Noto Sans Gunjala Gondi",
74
+ "Noto Sans Gurmukhi",
75
+ "Noto Sans Gurmukhi UI",
76
+ "Noto Sans Hanifi Rohingya",
77
+ "Noto Sans Hanunoo",
78
+ "Noto Sans Hatran",
79
+ "Noto Sans Hebrew",
80
+ "Noto Sans Hebrew Droid",
81
+ "Noto Sans Imperial Aramaic",
82
+ "Noto Sans Indic Siyaq Numbers",
83
+ "Noto Sans Inscriptional Pahlavi",
84
+ "Noto Sans Inscriptional Parthian",
85
+ "Noto Sans Javanese",
86
+ "Noto Sans Kaithi",
87
+ "Noto Sans Kannada",
88
+ "Noto Sans Kannada UI",
89
+ "Noto Sans Kayah Li",
90
+ "Noto Sans Kharoshthi",
91
+ "Noto Sans Khmer",
92
+ "Noto Sans Khmer UI",
93
+ "Noto Sans Khudawadi",
94
+ "Noto Sans Lao",
95
+ "Noto Sans Lao UI",
96
+ "Noto Sans Lepcha",
97
+ "Noto Sans Limbu",
98
+ "Noto Sans Linear A",
99
+ "Noto Sans Linear B",
100
+ "Noto Sans Lisu",
101
+ "Noto Sans Lycian",
102
+ "Noto Sans Lydian",
103
+ "Noto Sans Mahajani",
104
+ "Noto Sans Malayalam",
105
+ "Noto Sans Malayalam UI",
106
+ "Noto Sans Mandaic",
107
+ "Noto Sans Manichaean",
108
+ "Noto Sans Marchen",
109
+ "Noto Sans Masaram Gondi",
110
+ "Noto Sans Math",
111
+ "Noto Sans Mayan Numerals",
112
+ "Noto Sans Medefaidrin",
113
+ "Noto Sans Meetei Mayek",
114
+ "Noto Sans Mende Kikakui",
115
+ "Noto Sans Meroitic",
116
+ "Noto Sans Miao",
117
+ "Noto Sans Modi",
118
+ "Noto Sans Mongolian",
119
+ "Noto Sans Mono",
120
+ "Noto Sans Mro",
121
+ "Noto Sans Multani",
122
+ "Noto Sans Myanmar",
123
+ "Noto Sans Myanmar UI",
124
+ "Noto Sans NKo",
125
+ "Noto Sans Nabataean",
126
+ "Noto Sans Nandinagari",
127
+ "Noto Sans New Tai Lue",
128
+ "Noto Sans Newa",
129
+ "Noto Sans Nushu",
130
+ "Noto Sans Ogham",
131
+ "Noto Sans Old",
132
+ "Noto Sans Old Hungarian",
133
+ "Noto Sans Old North Arabian",
134
+ "Noto Sans Old Permic",
135
+ "Noto Sans Old Persian",
136
+ "Noto Sans Old Sogdian",
137
+ "Noto Sans Old South Arabian",
138
+ "Noto Sans Old Turkic",
139
+ "Noto Sans Oriya",
140
+ "Noto Sans Osage",
141
+ "Noto Sans Osmanya",
142
+ "Noto Sans Pahawh Hmong",
143
+ "Noto Sans Palmyrene",
144
+ "Noto Sans Pau Cin Hau",
145
+ "Noto Sans PhagsPa",
146
+ "Noto Sans Phoenician",
147
+ "Noto Sans Psalter Pahlavi",
148
+ "Noto Sans Rejang",
149
+ "Noto Sans Runic",
150
+ "Noto Sans Samaritan",
151
+ "Noto Sans Saurashtra",
152
+ "Noto Sans Sharada",
153
+ "Noto Sans Shavian",
154
+ "Noto Sans Siddham",
155
+ "Noto Sans SignWriting",
156
+ "Noto Sans Sinhala",
157
+ "Noto Sans Sinhala UI",
158
+ "Noto Sans Sogdian",
159
+ "Noto Sans Soyombo",
160
+ "Noto Sans Syloti Nagri",
161
+ "Noto Sans Symbols",
162
+ "Noto Sans Symbols2",
163
+ "Noto Sans Syriac",
164
+ "Noto Sans Tagalog",
165
+ "Noto Sans Tagbanwa",
166
+ "Noto Sans Tai Le",
167
+ "Noto Sans Tai Tham",
168
+ "Noto Sans Tai Viet",
169
+ "Noto Sans Takri",
170
+ "Noto Sans Tamil",
171
+ "Noto Sans Tamil Supplement",
172
+ "Noto Sans Tamil UI",
173
+ "Noto Sans Tangsa",
174
+ "Noto Sans Telugu",
175
+ "Noto Sans Telugu UI",
176
+ "Noto Sans Test",
177
+ "Noto Sans Thaana",
178
+ "Noto Sans Thai",
179
+ "Noto Sans Thai Looped",
180
+ "Noto Sans Thai Looped UI",
181
+ "Noto Sans Thai UI",
182
+ "Noto Sans Tifinagh",
183
+ "Noto Sans Tifinagh APT",
184
+ "Noto Sans Tifinagh Adrar",
185
+ "Noto Sans Tifinagh Agraw Imazighen",
186
+ "Noto Sans Tifinagh Ahaggar",
187
+ "Noto Sans Tifinagh Air",
188
+ "Noto Sans Tifinagh Azawagh",
189
+ "Noto Sans Tifinagh Ghat",
190
+ "Noto Sans Tifinagh Hawad",
191
+ "Noto Sans Tifinagh Rhissa Ixa",
192
+ "Noto Sans Tifinagh SIL",
193
+ "Noto Sans Tifinagh Tawellemmet",
194
+ "Noto Sans Tirhuta",
195
+ "Noto Sans Ugaritic",
196
+ "Noto Sans Vai",
197
+ "Noto Sans Vithkuqi",
198
+ "Noto Sans Wancho",
199
+ "Noto Sans Warang Citi",
200
+ "Noto Sans Yi",
201
+ "Noto Sans Zanabazar Square",
202
+ "Noto Traditional Nushu",
203
+ "P052",
204
+ "Sans",
205
+ "Serif",
206
+ "Source Code Pro",
207
+ "System-ui",
208
+ "URW Bookman",
209
+ "URW Gothic",
210
+ "Ubuntu",
211
+ "Ubuntu Condensed",
212
+ "Ubuntu Mono",
213
+ "Z003",
214
+ "Go Noto Current"
215
+ ],
216
+ "max_seq_length": 529,
217
+ "non_monospace_dpi": 120,
218
+ "non_monospace_font_file": "/scratch/project/dd-22-48/code/monospace_experiments_scripts/monospace_renderer/GoNotoCurrent.ttf",
219
+ "non_monospace_font_size": 8,
220
+ "pad_size": 3,
221
+ "pixels_per_patch": 16,
222
+ "rgb": true,
223
+ "text_renderer_type": "MonoPangoCairoTextRenderer"
224
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9836cbf6995d22ed625995c395e0a3b185cdcffe7981eeaf2436d50a2a13d201
3
+ size 3375