marinone94
commited on
Commit
β’
9eafbd0
1
Parent(s):
6b0ae6d
Training in progress, step 3000
Browse files- {checkpoint-768 β checkpoint-3000}/config.json +0 -0
- {checkpoint-768 β checkpoint-3000}/optimizer.pt +1 -1
- {checkpoint-768 β checkpoint-3000}/preprocessor_config.json +0 -0
- {checkpoint-768 β checkpoint-3000}/pytorch_model.bin +1 -1
- {checkpoint-768 β checkpoint-3000}/rng_state.pth +1 -1
- {checkpoint-768 β checkpoint-3000}/scaler.pt +1 -1
- {checkpoint-768 β checkpoint-3000}/scheduler.pt +1 -1
- checkpoint-3000/trainer_state.json +223 -0
- {checkpoint-768 β checkpoint-3000}/training_args.bin +1 -1
- checkpoint-768/trainer_state.json +0 -130
- pytorch_model.bin +1 -1
{checkpoint-768 β checkpoint-3000}/config.json
RENAMED
File without changes
|
{checkpoint-768 β checkpoint-3000}/optimizer.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 2490337809
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:94450a47c3ca62b991db9a57550d81b8cd90698ed4ef2543b783aa61e5c1766d
|
3 |
size 2490337809
|
{checkpoint-768 β checkpoint-3000}/preprocessor_config.json
RENAMED
File without changes
|
{checkpoint-768 β checkpoint-3000}/pytorch_model.bin
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1262063089
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:327bed48381999df4748f541a4a9fa0c348c7ae5d4e4f883c614432b3a8395bd
|
3 |
size 1262063089
|
{checkpoint-768 β checkpoint-3000}/rng_state.pth
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14567
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f80481468e633ed17f4d3b953b5bfd7af41f1fc73703fdad42f6fc3bad5c020f
|
3 |
size 14567
|
{checkpoint-768 β checkpoint-3000}/scaler.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 559
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:05430fc7d2676bfa8c337884b4956f58b83fbf1ce211afb4952306653777fc6c
|
3 |
size 559
|
{checkpoint-768 β checkpoint-3000}/scheduler.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 623
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:14f025f6209560d43d4588b308395c8f0162d863f3099cf81ce8917d61699085
|
3 |
size 623
|
checkpoint-3000/trainer_state.json
ADDED
@@ -0,0 +1,223 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 31.24935064935065,
|
5 |
+
"global_step": 3000,
|
6 |
+
"is_hyper_param_search": false,
|
7 |
+
"is_local_process_zero": true,
|
8 |
+
"is_world_process_zero": true,
|
9 |
+
"log_history": [
|
10 |
+
{
|
11 |
+
"epoch": 1.04,
|
12 |
+
"learning_rate": 3.828125e-06,
|
13 |
+
"loss": 15.104,
|
14 |
+
"step": 100
|
15 |
+
},
|
16 |
+
{
|
17 |
+
"epoch": 2.08,
|
18 |
+
"learning_rate": 7.6953125e-06,
|
19 |
+
"loss": 12.2149,
|
20 |
+
"step": 200
|
21 |
+
},
|
22 |
+
{
|
23 |
+
"epoch": 3.12,
|
24 |
+
"learning_rate": 1.16015625e-05,
|
25 |
+
"loss": 4.9885,
|
26 |
+
"step": 300
|
27 |
+
},
|
28 |
+
{
|
29 |
+
"epoch": 4.17,
|
30 |
+
"learning_rate": 1.55078125e-05,
|
31 |
+
"loss": 3.3788,
|
32 |
+
"step": 400
|
33 |
+
},
|
34 |
+
{
|
35 |
+
"epoch": 5.21,
|
36 |
+
"learning_rate": 1.94140625e-05,
|
37 |
+
"loss": 3.1705,
|
38 |
+
"step": 500
|
39 |
+
},
|
40 |
+
{
|
41 |
+
"epoch": 6.25,
|
42 |
+
"learning_rate": 2.3320312499999995e-05,
|
43 |
+
"loss": 3.11,
|
44 |
+
"step": 600
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 7.29,
|
48 |
+
"learning_rate": 2.72265625e-05,
|
49 |
+
"loss": 3.0526,
|
50 |
+
"step": 700
|
51 |
+
},
|
52 |
+
{
|
53 |
+
"epoch": 8.33,
|
54 |
+
"learning_rate": 3.11328125e-05,
|
55 |
+
"loss": 3.0113,
|
56 |
+
"step": 800
|
57 |
+
},
|
58 |
+
{
|
59 |
+
"epoch": 9.37,
|
60 |
+
"learning_rate": 3.5039062499999995e-05,
|
61 |
+
"loss": 2.9601,
|
62 |
+
"step": 900
|
63 |
+
},
|
64 |
+
{
|
65 |
+
"epoch": 10.42,
|
66 |
+
"learning_rate": 3.89453125e-05,
|
67 |
+
"loss": 2.9099,
|
68 |
+
"step": 1000
|
69 |
+
},
|
70 |
+
{
|
71 |
+
"epoch": 10.42,
|
72 |
+
"eval_loss": 2.8369038105010986,
|
73 |
+
"eval_runtime": 198.5866,
|
74 |
+
"eval_samples_per_second": 25.405,
|
75 |
+
"eval_steps_per_second": 0.796,
|
76 |
+
"eval_wer": 1.0,
|
77 |
+
"step": 1000
|
78 |
+
},
|
79 |
+
{
|
80 |
+
"epoch": 11.46,
|
81 |
+
"learning_rate": 4.28515625e-05,
|
82 |
+
"loss": 2.6653,
|
83 |
+
"step": 1100
|
84 |
+
},
|
85 |
+
{
|
86 |
+
"epoch": 12.5,
|
87 |
+
"learning_rate": 4.675781249999999e-05,
|
88 |
+
"loss": 1.8958,
|
89 |
+
"step": 1200
|
90 |
+
},
|
91 |
+
{
|
92 |
+
"epoch": 13.54,
|
93 |
+
"learning_rate": 5.0664062499999996e-05,
|
94 |
+
"loss": 1.4812,
|
95 |
+
"step": 1300
|
96 |
+
},
|
97 |
+
{
|
98 |
+
"epoch": 14.58,
|
99 |
+
"learning_rate": 5.45703125e-05,
|
100 |
+
"loss": 1.3358,
|
101 |
+
"step": 1400
|
102 |
+
},
|
103 |
+
{
|
104 |
+
"epoch": 15.62,
|
105 |
+
"learning_rate": 5.8476562499999996e-05,
|
106 |
+
"loss": 1.2522,
|
107 |
+
"step": 1500
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 16.66,
|
111 |
+
"learning_rate": 6.238281249999999e-05,
|
112 |
+
"loss": 1.1919,
|
113 |
+
"step": 1600
|
114 |
+
},
|
115 |
+
{
|
116 |
+
"epoch": 17.71,
|
117 |
+
"learning_rate": 6.62890625e-05,
|
118 |
+
"loss": 1.1663,
|
119 |
+
"step": 1700
|
120 |
+
},
|
121 |
+
{
|
122 |
+
"epoch": 18.75,
|
123 |
+
"learning_rate": 7.019531249999999e-05,
|
124 |
+
"loss": 1.1289,
|
125 |
+
"step": 1800
|
126 |
+
},
|
127 |
+
{
|
128 |
+
"epoch": 19.79,
|
129 |
+
"learning_rate": 7.410156249999999e-05,
|
130 |
+
"loss": 1.0902,
|
131 |
+
"step": 1900
|
132 |
+
},
|
133 |
+
{
|
134 |
+
"epoch": 20.83,
|
135 |
+
"learning_rate": 7.4248046875e-05,
|
136 |
+
"loss": 1.0745,
|
137 |
+
"step": 2000
|
138 |
+
},
|
139 |
+
{
|
140 |
+
"epoch": 20.83,
|
141 |
+
"eval_loss": 0.19572903215885162,
|
142 |
+
"eval_runtime": 202.06,
|
143 |
+
"eval_samples_per_second": 24.968,
|
144 |
+
"eval_steps_per_second": 0.782,
|
145 |
+
"eval_wer": 0.16725371193237237,
|
146 |
+
"step": 2000
|
147 |
+
},
|
148 |
+
{
|
149 |
+
"epoch": 21.87,
|
150 |
+
"learning_rate": 7.327148437499999e-05,
|
151 |
+
"loss": 1.0485,
|
152 |
+
"step": 2100
|
153 |
+
},
|
154 |
+
{
|
155 |
+
"epoch": 22.91,
|
156 |
+
"learning_rate": 7.2294921875e-05,
|
157 |
+
"loss": 1.0291,
|
158 |
+
"step": 2200
|
159 |
+
},
|
160 |
+
{
|
161 |
+
"epoch": 23.96,
|
162 |
+
"learning_rate": 7.1318359375e-05,
|
163 |
+
"loss": 1.007,
|
164 |
+
"step": 2300
|
165 |
+
},
|
166 |
+
{
|
167 |
+
"epoch": 25.0,
|
168 |
+
"learning_rate": 7.0341796875e-05,
|
169 |
+
"loss": 1.0008,
|
170 |
+
"step": 2400
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 26.04,
|
174 |
+
"learning_rate": 6.9365234375e-05,
|
175 |
+
"loss": 0.988,
|
176 |
+
"step": 2500
|
177 |
+
},
|
178 |
+
{
|
179 |
+
"epoch": 27.08,
|
180 |
+
"learning_rate": 6.8388671875e-05,
|
181 |
+
"loss": 0.9766,
|
182 |
+
"step": 2600
|
183 |
+
},
|
184 |
+
{
|
185 |
+
"epoch": 28.12,
|
186 |
+
"learning_rate": 6.7412109375e-05,
|
187 |
+
"loss": 0.9663,
|
188 |
+
"step": 2700
|
189 |
+
},
|
190 |
+
{
|
191 |
+
"epoch": 29.17,
|
192 |
+
"learning_rate": 6.6435546875e-05,
|
193 |
+
"loss": 0.9539,
|
194 |
+
"step": 2800
|
195 |
+
},
|
196 |
+
{
|
197 |
+
"epoch": 30.21,
|
198 |
+
"learning_rate": 6.545898437499999e-05,
|
199 |
+
"loss": 0.9479,
|
200 |
+
"step": 2900
|
201 |
+
},
|
202 |
+
{
|
203 |
+
"epoch": 31.25,
|
204 |
+
"learning_rate": 6.448242187499999e-05,
|
205 |
+
"loss": 0.934,
|
206 |
+
"step": 3000
|
207 |
+
},
|
208 |
+
{
|
209 |
+
"epoch": 31.25,
|
210 |
+
"eval_loss": 0.1579357087612152,
|
211 |
+
"eval_runtime": 198.7733,
|
212 |
+
"eval_samples_per_second": 25.381,
|
213 |
+
"eval_steps_per_second": 0.795,
|
214 |
+
"eval_wer": 0.1388858784003468,
|
215 |
+
"step": 3000
|
216 |
+
}
|
217 |
+
],
|
218 |
+
"max_steps": 9600,
|
219 |
+
"num_train_epochs": 100,
|
220 |
+
"total_flos": 4.660374091191037e+19,
|
221 |
+
"trial_name": null,
|
222 |
+
"trial_params": null
|
223 |
+
}
|
{checkpoint-768 β checkpoint-3000}/training_args.bin
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 3055
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:28f8dab6838e0cc8a253bbc1fb16c34a7a36619cf3b9e2bd9c1e4027635a3b2b
|
3 |
size 3055
|
checkpoint-768/trainer_state.json
DELETED
@@ -1,130 +0,0 @@
|
|
1 |
-
{
|
2 |
-
"best_metric": null,
|
3 |
-
"best_model_checkpoint": null,
|
4 |
-
"epoch": 7.997402597402598,
|
5 |
-
"global_step": 768,
|
6 |
-
"is_hyper_param_search": false,
|
7 |
-
"is_local_process_zero": true,
|
8 |
-
"is_world_process_zero": true,
|
9 |
-
"log_history": [
|
10 |
-
{
|
11 |
-
"epoch": 1.0,
|
12 |
-
"eval_loss": 18.22025489807129,
|
13 |
-
"eval_runtime": 188.3513,
|
14 |
-
"eval_samples_per_second": 26.785,
|
15 |
-
"eval_steps_per_second": 0.839,
|
16 |
-
"eval_wer": 1.0021133629565406,
|
17 |
-
"step": 96
|
18 |
-
},
|
19 |
-
{
|
20 |
-
"epoch": 1.04,
|
21 |
-
"learning_rate": 1.5156249999999998e-06,
|
22 |
-
"loss": 15.219,
|
23 |
-
"step": 100
|
24 |
-
},
|
25 |
-
{
|
26 |
-
"epoch": 2.0,
|
27 |
-
"eval_loss": 17.49703025817871,
|
28 |
-
"eval_runtime": 186.0672,
|
29 |
-
"eval_samples_per_second": 27.114,
|
30 |
-
"eval_steps_per_second": 0.849,
|
31 |
-
"eval_wer": 1.0,
|
32 |
-
"step": 192
|
33 |
-
},
|
34 |
-
{
|
35 |
-
"epoch": 2.08,
|
36 |
-
"learning_rate": 3.0781249999999994e-06,
|
37 |
-
"loss": 14.7046,
|
38 |
-
"step": 200
|
39 |
-
},
|
40 |
-
{
|
41 |
-
"epoch": 3.0,
|
42 |
-
"eval_loss": 11.056963920593262,
|
43 |
-
"eval_runtime": 184.9586,
|
44 |
-
"eval_samples_per_second": 27.276,
|
45 |
-
"eval_steps_per_second": 0.854,
|
46 |
-
"eval_wer": 1.0,
|
47 |
-
"step": 288
|
48 |
-
},
|
49 |
-
{
|
50 |
-
"epoch": 3.12,
|
51 |
-
"learning_rate": 4.640624999999999e-06,
|
52 |
-
"loss": 11.6089,
|
53 |
-
"step": 300
|
54 |
-
},
|
55 |
-
{
|
56 |
-
"epoch": 4.0,
|
57 |
-
"eval_loss": 5.172672271728516,
|
58 |
-
"eval_runtime": 185.1159,
|
59 |
-
"eval_samples_per_second": 27.253,
|
60 |
-
"eval_steps_per_second": 0.854,
|
61 |
-
"eval_wer": 1.0,
|
62 |
-
"step": 384
|
63 |
-
},
|
64 |
-
{
|
65 |
-
"epoch": 4.17,
|
66 |
-
"learning_rate": 6.203124999999999e-06,
|
67 |
-
"loss": 5.9408,
|
68 |
-
"step": 400
|
69 |
-
},
|
70 |
-
{
|
71 |
-
"epoch": 5.0,
|
72 |
-
"eval_loss": 3.6944665908813477,
|
73 |
-
"eval_runtime": 186.0698,
|
74 |
-
"eval_samples_per_second": 27.113,
|
75 |
-
"eval_steps_per_second": 0.849,
|
76 |
-
"eval_wer": 1.0,
|
77 |
-
"step": 480
|
78 |
-
},
|
79 |
-
{
|
80 |
-
"epoch": 5.21,
|
81 |
-
"learning_rate": 7.765624999999999e-06,
|
82 |
-
"loss": 3.8196,
|
83 |
-
"step": 500
|
84 |
-
},
|
85 |
-
{
|
86 |
-
"epoch": 6.0,
|
87 |
-
"eval_loss": 3.3380837440490723,
|
88 |
-
"eval_runtime": 183.9322,
|
89 |
-
"eval_samples_per_second": 27.429,
|
90 |
-
"eval_steps_per_second": 0.859,
|
91 |
-
"eval_wer": 1.0,
|
92 |
-
"step": 576
|
93 |
-
},
|
94 |
-
{
|
95 |
-
"epoch": 6.25,
|
96 |
-
"learning_rate": 9.328125e-06,
|
97 |
-
"loss": 3.3535,
|
98 |
-
"step": 600
|
99 |
-
},
|
100 |
-
{
|
101 |
-
"epoch": 7.0,
|
102 |
-
"eval_loss": 3.2117931842803955,
|
103 |
-
"eval_runtime": 186.1,
|
104 |
-
"eval_samples_per_second": 27.109,
|
105 |
-
"eval_steps_per_second": 0.849,
|
106 |
-
"eval_wer": 1.0,
|
107 |
-
"step": 672
|
108 |
-
},
|
109 |
-
{
|
110 |
-
"epoch": 7.29,
|
111 |
-
"learning_rate": 1.0890624999999999e-05,
|
112 |
-
"loss": 3.1949,
|
113 |
-
"step": 700
|
114 |
-
},
|
115 |
-
{
|
116 |
-
"epoch": 8.0,
|
117 |
-
"eval_loss": 3.158613920211792,
|
118 |
-
"eval_runtime": 182.4287,
|
119 |
-
"eval_samples_per_second": 27.655,
|
120 |
-
"eval_steps_per_second": 0.866,
|
121 |
-
"eval_wer": 1.0,
|
122 |
-
"step": 768
|
123 |
-
}
|
124 |
-
],
|
125 |
-
"max_steps": 19200,
|
126 |
-
"num_train_epochs": 200,
|
127 |
-
"total_flos": 1.1926542720934052e+19,
|
128 |
-
"trial_name": null,
|
129 |
-
"trial_params": null
|
130 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1262063089
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:327bed48381999df4748f541a4a9fa0c348c7ae5d4e4f883c614432b3a8395bd
|
3 |
size 1262063089
|