add tokenizer
Browse files- checkpoint-1500/config.json +0 -109
- checkpoint-1500/optimizer.pt +0 -3
- checkpoint-1500/preprocessor_config.json +0 -9
- checkpoint-1500/pytorch_model.bin +0 -3
- checkpoint-1500/rng_state.pth +0 -3
- checkpoint-1500/scaler.pt +0 -3
- checkpoint-1500/scheduler.pt +0 -3
- checkpoint-1500/trainer_state.json +0 -1843
- checkpoint-1500/training_args.bin +0 -3
- checkpoint-2000/config.json +0 -109
- checkpoint-2000/optimizer.pt +0 -3
- checkpoint-2000/preprocessor_config.json +0 -9
- checkpoint-2000/pytorch_model.bin +0 -3
- checkpoint-2000/rng_state.pth +0 -3
- checkpoint-2000/scaler.pt +0 -3
- checkpoint-2000/scheduler.pt +0 -3
- checkpoint-2000/trainer_state.json +0 -2452
- checkpoint-2000/training_args.bin +0 -3
checkpoint-1500/config.json
DELETED
@@ -1,109 +0,0 @@
|
|
1 |
-
{
|
2 |
-
"_name_or_path": "facebook/wav2vec2-base-it-voxpopuli",
|
3 |
-
"activation_dropout": 0.1,
|
4 |
-
"adapter_kernel_size": 3,
|
5 |
-
"adapter_stride": 2,
|
6 |
-
"add_adapter": false,
|
7 |
-
"apply_spec_augment": true,
|
8 |
-
"architectures": [
|
9 |
-
"Wav2Vec2ForCTC"
|
10 |
-
],
|
11 |
-
"attention_dropout": 0.1,
|
12 |
-
"bos_token_id": 1,
|
13 |
-
"classifier_proj_size": 256,
|
14 |
-
"codevector_dim": 256,
|
15 |
-
"contrastive_logits_temperature": 0.1,
|
16 |
-
"conv_bias": false,
|
17 |
-
"conv_dim": [
|
18 |
-
512,
|
19 |
-
512,
|
20 |
-
512,
|
21 |
-
512,
|
22 |
-
512,
|
23 |
-
512,
|
24 |
-
512
|
25 |
-
],
|
26 |
-
"conv_kernel": [
|
27 |
-
10,
|
28 |
-
3,
|
29 |
-
3,
|
30 |
-
3,
|
31 |
-
3,
|
32 |
-
2,
|
33 |
-
2
|
34 |
-
],
|
35 |
-
"conv_stride": [
|
36 |
-
5,
|
37 |
-
2,
|
38 |
-
2,
|
39 |
-
2,
|
40 |
-
2,
|
41 |
-
2,
|
42 |
-
2
|
43 |
-
],
|
44 |
-
"ctc_loss_reduction": "mean",
|
45 |
-
"ctc_zero_infinity": false,
|
46 |
-
"diversity_loss_weight": 0.1,
|
47 |
-
"do_stable_layer_norm": false,
|
48 |
-
"eos_token_id": 2,
|
49 |
-
"feat_extract_activation": "gelu",
|
50 |
-
"feat_extract_dropout": 0.0,
|
51 |
-
"feat_extract_norm": "group",
|
52 |
-
"feat_proj_dropout": 0.1,
|
53 |
-
"feat_quantizer_dropout": 0.0,
|
54 |
-
"final_dropout": 0.1,
|
55 |
-
"gradient_checkpointing": false,
|
56 |
-
"hidden_act": "gelu",
|
57 |
-
"hidden_dropout": 0.1,
|
58 |
-
"hidden_dropout_prob": 0.1,
|
59 |
-
"hidden_size": 768,
|
60 |
-
"initializer_range": 0.02,
|
61 |
-
"intermediate_size": 3072,
|
62 |
-
"layer_norm_eps": 1e-05,
|
63 |
-
"layerdrop": 0.1,
|
64 |
-
"mask_feature_length": 10,
|
65 |
-
"mask_feature_min_masks": 0,
|
66 |
-
"mask_feature_prob": 0.0,
|
67 |
-
"mask_time_length": 10,
|
68 |
-
"mask_time_min_masks": 2,
|
69 |
-
"mask_time_prob": 0.05,
|
70 |
-
"model_type": "wav2vec2",
|
71 |
-
"num_adapter_layers": 3,
|
72 |
-
"num_attention_heads": 12,
|
73 |
-
"num_codevector_groups": 2,
|
74 |
-
"num_codevectors_per_group": 320,
|
75 |
-
"num_conv_pos_embedding_groups": 16,
|
76 |
-
"num_conv_pos_embeddings": 128,
|
77 |
-
"num_feat_extract_layers": 7,
|
78 |
-
"num_hidden_layers": 12,
|
79 |
-
"num_negatives": 100,
|
80 |
-
"output_hidden_size": 768,
|
81 |
-
"pad_token_id": 25,
|
82 |
-
"proj_codevector_dim": 256,
|
83 |
-
"tdnn_dilation": [
|
84 |
-
1,
|
85 |
-
2,
|
86 |
-
3,
|
87 |
-
1,
|
88 |
-
1
|
89 |
-
],
|
90 |
-
"tdnn_dim": [
|
91 |
-
512,
|
92 |
-
512,
|
93 |
-
512,
|
94 |
-
512,
|
95 |
-
1500
|
96 |
-
],
|
97 |
-
"tdnn_kernel": [
|
98 |
-
5,
|
99 |
-
3,
|
100 |
-
3,
|
101 |
-
1,
|
102 |
-
1
|
103 |
-
],
|
104 |
-
"torch_dtype": "float32",
|
105 |
-
"transformers_version": "4.16.1",
|
106 |
-
"use_weighted_layer_sum": false,
|
107 |
-
"vocab_size": 32,
|
108 |
-
"xvector_output_dim": 512
|
109 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-1500/optimizer.pt
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:09193c765095b0edd50bbf764532b0a78939a909d4dccdfd6eeea5ecfa2d02dc
|
3 |
-
size 721685265
|
|
|
|
|
|
|
|
checkpoint-1500/preprocessor_config.json
DELETED
@@ -1,9 +0,0 @@
|
|
1 |
-
{
|
2 |
-
"do_normalize": true,
|
3 |
-
"feature_extractor_type": "Wav2Vec2FeatureExtractor",
|
4 |
-
"feature_size": 1,
|
5 |
-
"padding_side": "right",
|
6 |
-
"padding_value": 0.0,
|
7 |
-
"return_attention_mask": false,
|
8 |
-
"sampling_rate": 16000
|
9 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-1500/pytorch_model.bin
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:9b7f06101e1e8d6a711b5db6d76ed9a5bc122a74a8cc304c9d31fe46f693d853
|
3 |
-
size 377670039
|
|
|
|
|
|
|
|
checkpoint-1500/rng_state.pth
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:1531a7e0f905290b8961b8267ec7ef6cc83660736b9a4acaec582bf5f0e57520
|
3 |
-
size 14567
|
|
|
|
|
|
|
|
checkpoint-1500/scaler.pt
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:ffce1add4731473c42f90f9591511f4cbeeff8f9e059b22cd53faea5621924cb
|
3 |
-
size 559
|
|
|
|
|
|
|
|
checkpoint-1500/scheduler.pt
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:d212565847c568a9d9ec141c2168db2ea3ae4a7aab188a7409e99ecfd29bf6a1
|
3 |
-
size 623
|
|
|
|
|
|
|
|
checkpoint-1500/trainer_state.json
DELETED
@@ -1,1843 +0,0 @@
|
|
1 |
-
{
|
2 |
-
"best_metric": null,
|
3 |
-
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0.4191114836546521,
|
5 |
-
"global_step": 1500,
|
6 |
-
"is_hyper_param_search": false,
|
7 |
-
"is_local_process_zero": true,
|
8 |
-
"is_world_process_zero": true,
|
9 |
-
"log_history": [
|
10 |
-
{
|
11 |
-
"epoch": 0.0,
|
12 |
-
"learning_rate": 1e-08,
|
13 |
-
"loss": 24.3697,
|
14 |
-
"step": 5
|
15 |
-
},
|
16 |
-
{
|
17 |
-
"epoch": 0.0,
|
18 |
-
"learning_rate": 3.0000000000000004e-08,
|
19 |
-
"loss": 17.6279,
|
20 |
-
"step": 10
|
21 |
-
},
|
22 |
-
{
|
23 |
-
"epoch": 0.0,
|
24 |
-
"learning_rate": 5.5e-08,
|
25 |
-
"loss": 17.3553,
|
26 |
-
"step": 15
|
27 |
-
},
|
28 |
-
{
|
29 |
-
"epoch": 0.01,
|
30 |
-
"learning_rate": 8e-08,
|
31 |
-
"loss": 16.801,
|
32 |
-
"step": 20
|
33 |
-
},
|
34 |
-
{
|
35 |
-
"epoch": 0.01,
|
36 |
-
"learning_rate": 1.0500000000000001e-07,
|
37 |
-
"loss": 15.1858,
|
38 |
-
"step": 25
|
39 |
-
},
|
40 |
-
{
|
41 |
-
"epoch": 0.01,
|
42 |
-
"learning_rate": 1.3e-07,
|
43 |
-
"loss": 14.8524,
|
44 |
-
"step": 30
|
45 |
-
},
|
46 |
-
{
|
47 |
-
"epoch": 0.01,
|
48 |
-
"learning_rate": 1.5500000000000002e-07,
|
49 |
-
"loss": 14.8696,
|
50 |
-
"step": 35
|
51 |
-
},
|
52 |
-
{
|
53 |
-
"epoch": 0.01,
|
54 |
-
"learning_rate": 1.8e-07,
|
55 |
-
"loss": 14.5785,
|
56 |
-
"step": 40
|
57 |
-
},
|
58 |
-
{
|
59 |
-
"epoch": 0.01,
|
60 |
-
"learning_rate": 2.0500000000000002e-07,
|
61 |
-
"loss": 15.0479,
|
62 |
-
"step": 45
|
63 |
-
},
|
64 |
-
{
|
65 |
-
"epoch": 0.01,
|
66 |
-
"learning_rate": 2.3000000000000002e-07,
|
67 |
-
"loss": 14.5691,
|
68 |
-
"step": 50
|
69 |
-
},
|
70 |
-
{
|
71 |
-
"epoch": 0.02,
|
72 |
-
"learning_rate": 2.55e-07,
|
73 |
-
"loss": 18.4354,
|
74 |
-
"step": 55
|
75 |
-
},
|
76 |
-
{
|
77 |
-
"epoch": 0.02,
|
78 |
-
"learning_rate": 2.8e-07,
|
79 |
-
"loss": 18.1091,
|
80 |
-
"step": 60
|
81 |
-
},
|
82 |
-
{
|
83 |
-
"epoch": 0.02,
|
84 |
-
"learning_rate": 3.0500000000000004e-07,
|
85 |
-
"loss": 17.4985,
|
86 |
-
"step": 65
|
87 |
-
},
|
88 |
-
{
|
89 |
-
"epoch": 0.02,
|
90 |
-
"learning_rate": 3.3e-07,
|
91 |
-
"loss": 15.3995,
|
92 |
-
"step": 70
|
93 |
-
},
|
94 |
-
{
|
95 |
-
"epoch": 0.02,
|
96 |
-
"learning_rate": 3.55e-07,
|
97 |
-
"loss": 15.2441,
|
98 |
-
"step": 75
|
99 |
-
},
|
100 |
-
{
|
101 |
-
"epoch": 0.02,
|
102 |
-
"learning_rate": 3.8e-07,
|
103 |
-
"loss": 14.5564,
|
104 |
-
"step": 80
|
105 |
-
},
|
106 |
-
{
|
107 |
-
"epoch": 0.02,
|
108 |
-
"learning_rate": 4.0500000000000004e-07,
|
109 |
-
"loss": 14.824,
|
110 |
-
"step": 85
|
111 |
-
},
|
112 |
-
{
|
113 |
-
"epoch": 0.03,
|
114 |
-
"learning_rate": 4.3e-07,
|
115 |
-
"loss": 14.9045,
|
116 |
-
"step": 90
|
117 |
-
},
|
118 |
-
{
|
119 |
-
"epoch": 0.03,
|
120 |
-
"learning_rate": 4.5500000000000004e-07,
|
121 |
-
"loss": 14.4084,
|
122 |
-
"step": 95
|
123 |
-
},
|
124 |
-
{
|
125 |
-
"epoch": 0.03,
|
126 |
-
"learning_rate": 4.800000000000001e-07,
|
127 |
-
"loss": 14.5942,
|
128 |
-
"step": 100
|
129 |
-
},
|
130 |
-
{
|
131 |
-
"epoch": 0.03,
|
132 |
-
"learning_rate": 5.05e-07,
|
133 |
-
"loss": 18.0525,
|
134 |
-
"step": 105
|
135 |
-
},
|
136 |
-
{
|
137 |
-
"epoch": 0.03,
|
138 |
-
"learning_rate": 5.3e-07,
|
139 |
-
"loss": 16.5825,
|
140 |
-
"step": 110
|
141 |
-
},
|
142 |
-
{
|
143 |
-
"epoch": 0.03,
|
144 |
-
"learning_rate": 5.550000000000001e-07,
|
145 |
-
"loss": 17.0772,
|
146 |
-
"step": 115
|
147 |
-
},
|
148 |
-
{
|
149 |
-
"epoch": 0.03,
|
150 |
-
"learning_rate": 5.800000000000001e-07,
|
151 |
-
"loss": 15.6476,
|
152 |
-
"step": 120
|
153 |
-
},
|
154 |
-
{
|
155 |
-
"epoch": 0.03,
|
156 |
-
"learning_rate": 6.05e-07,
|
157 |
-
"loss": 14.2361,
|
158 |
-
"step": 125
|
159 |
-
},
|
160 |
-
{
|
161 |
-
"epoch": 0.04,
|
162 |
-
"learning_rate": 6.3e-07,
|
163 |
-
"loss": 14.3617,
|
164 |
-
"step": 130
|
165 |
-
},
|
166 |
-
{
|
167 |
-
"epoch": 0.04,
|
168 |
-
"learning_rate": 6.550000000000001e-07,
|
169 |
-
"loss": 14.0812,
|
170 |
-
"step": 135
|
171 |
-
},
|
172 |
-
{
|
173 |
-
"epoch": 0.04,
|
174 |
-
"learning_rate": 6.800000000000001e-07,
|
175 |
-
"loss": 13.7279,
|
176 |
-
"step": 140
|
177 |
-
},
|
178 |
-
{
|
179 |
-
"epoch": 0.04,
|
180 |
-
"learning_rate": 7.05e-07,
|
181 |
-
"loss": 13.5363,
|
182 |
-
"step": 145
|
183 |
-
},
|
184 |
-
{
|
185 |
-
"epoch": 0.04,
|
186 |
-
"learning_rate": 7.3e-07,
|
187 |
-
"loss": 14.8066,
|
188 |
-
"step": 150
|
189 |
-
},
|
190 |
-
{
|
191 |
-
"epoch": 0.04,
|
192 |
-
"learning_rate": 7.5e-07,
|
193 |
-
"loss": 17.2667,
|
194 |
-
"step": 155
|
195 |
-
},
|
196 |
-
{
|
197 |
-
"epoch": 0.04,
|
198 |
-
"learning_rate": 7.750000000000001e-07,
|
199 |
-
"loss": 15.1567,
|
200 |
-
"step": 160
|
201 |
-
},
|
202 |
-
{
|
203 |
-
"epoch": 0.05,
|
204 |
-
"learning_rate": 8.000000000000001e-07,
|
205 |
-
"loss": 16.3138,
|
206 |
-
"step": 165
|
207 |
-
},
|
208 |
-
{
|
209 |
-
"epoch": 0.05,
|
210 |
-
"learning_rate": 8.250000000000001e-07,
|
211 |
-
"loss": 13.1312,
|
212 |
-
"step": 170
|
213 |
-
},
|
214 |
-
{
|
215 |
-
"epoch": 0.05,
|
216 |
-
"learning_rate": 8.500000000000001e-07,
|
217 |
-
"loss": 12.7545,
|
218 |
-
"step": 175
|
219 |
-
},
|
220 |
-
{
|
221 |
-
"epoch": 0.05,
|
222 |
-
"learning_rate": 8.75e-07,
|
223 |
-
"loss": 13.0234,
|
224 |
-
"step": 180
|
225 |
-
},
|
226 |
-
{
|
227 |
-
"epoch": 0.05,
|
228 |
-
"learning_rate": 9.000000000000001e-07,
|
229 |
-
"loss": 13.2179,
|
230 |
-
"step": 185
|
231 |
-
},
|
232 |
-
{
|
233 |
-
"epoch": 0.05,
|
234 |
-
"learning_rate": 9.25e-07,
|
235 |
-
"loss": 13.0916,
|
236 |
-
"step": 190
|
237 |
-
},
|
238 |
-
{
|
239 |
-
"epoch": 0.05,
|
240 |
-
"learning_rate": 9.500000000000001e-07,
|
241 |
-
"loss": 11.6765,
|
242 |
-
"step": 195
|
243 |
-
},
|
244 |
-
{
|
245 |
-
"epoch": 0.06,
|
246 |
-
"learning_rate": 9.750000000000002e-07,
|
247 |
-
"loss": 12.5044,
|
248 |
-
"step": 200
|
249 |
-
},
|
250 |
-
{
|
251 |
-
"epoch": 0.06,
|
252 |
-
"learning_rate": 1.0000000000000002e-06,
|
253 |
-
"loss": 15.0522,
|
254 |
-
"step": 205
|
255 |
-
},
|
256 |
-
{
|
257 |
-
"epoch": 0.06,
|
258 |
-
"learning_rate": 1.025e-06,
|
259 |
-
"loss": 12.4961,
|
260 |
-
"step": 210
|
261 |
-
},
|
262 |
-
{
|
263 |
-
"epoch": 0.06,
|
264 |
-
"learning_rate": 1.0500000000000001e-06,
|
265 |
-
"loss": 11.6473,
|
266 |
-
"step": 215
|
267 |
-
},
|
268 |
-
{
|
269 |
-
"epoch": 0.06,
|
270 |
-
"learning_rate": 1.075e-06,
|
271 |
-
"loss": 10.5303,
|
272 |
-
"step": 220
|
273 |
-
},
|
274 |
-
{
|
275 |
-
"epoch": 0.06,
|
276 |
-
"learning_rate": 1.1e-06,
|
277 |
-
"loss": 12.0959,
|
278 |
-
"step": 225
|
279 |
-
},
|
280 |
-
{
|
281 |
-
"epoch": 0.06,
|
282 |
-
"learning_rate": 1.125e-06,
|
283 |
-
"loss": 10.7053,
|
284 |
-
"step": 230
|
285 |
-
},
|
286 |
-
{
|
287 |
-
"epoch": 0.07,
|
288 |
-
"learning_rate": 1.1500000000000002e-06,
|
289 |
-
"loss": 9.5165,
|
290 |
-
"step": 235
|
291 |
-
},
|
292 |
-
{
|
293 |
-
"epoch": 0.07,
|
294 |
-
"learning_rate": 1.175e-06,
|
295 |
-
"loss": 11.838,
|
296 |
-
"step": 240
|
297 |
-
},
|
298 |
-
{
|
299 |
-
"epoch": 0.07,
|
300 |
-
"learning_rate": 1.2000000000000002e-06,
|
301 |
-
"loss": 9.5919,
|
302 |
-
"step": 245
|
303 |
-
},
|
304 |
-
{
|
305 |
-
"epoch": 0.07,
|
306 |
-
"learning_rate": 1.2250000000000001e-06,
|
307 |
-
"loss": 9.7433,
|
308 |
-
"step": 250
|
309 |
-
},
|
310 |
-
{
|
311 |
-
"epoch": 0.07,
|
312 |
-
"learning_rate": 1.25e-06,
|
313 |
-
"loss": 10.6129,
|
314 |
-
"step": 255
|
315 |
-
},
|
316 |
-
{
|
317 |
-
"epoch": 0.07,
|
318 |
-
"learning_rate": 1.275e-06,
|
319 |
-
"loss": 9.6936,
|
320 |
-
"step": 260
|
321 |
-
},
|
322 |
-
{
|
323 |
-
"epoch": 0.07,
|
324 |
-
"learning_rate": 1.3e-06,
|
325 |
-
"loss": 10.8679,
|
326 |
-
"step": 265
|
327 |
-
},
|
328 |
-
{
|
329 |
-
"epoch": 0.08,
|
330 |
-
"learning_rate": 1.3250000000000002e-06,
|
331 |
-
"loss": 12.5521,
|
332 |
-
"step": 270
|
333 |
-
},
|
334 |
-
{
|
335 |
-
"epoch": 0.08,
|
336 |
-
"learning_rate": 1.3500000000000002e-06,
|
337 |
-
"loss": 8.2902,
|
338 |
-
"step": 275
|
339 |
-
},
|
340 |
-
{
|
341 |
-
"epoch": 0.08,
|
342 |
-
"learning_rate": 1.3750000000000002e-06,
|
343 |
-
"loss": 8.7901,
|
344 |
-
"step": 280
|
345 |
-
},
|
346 |
-
{
|
347 |
-
"epoch": 0.08,
|
348 |
-
"learning_rate": 1.4000000000000001e-06,
|
349 |
-
"loss": 7.9864,
|
350 |
-
"step": 285
|
351 |
-
},
|
352 |
-
{
|
353 |
-
"epoch": 0.08,
|
354 |
-
"learning_rate": 1.425e-06,
|
355 |
-
"loss": 8.0575,
|
356 |
-
"step": 290
|
357 |
-
},
|
358 |
-
{
|
359 |
-
"epoch": 0.08,
|
360 |
-
"learning_rate": 1.45e-06,
|
361 |
-
"loss": 7.4836,
|
362 |
-
"step": 295
|
363 |
-
},
|
364 |
-
{
|
365 |
-
"epoch": 0.08,
|
366 |
-
"learning_rate": 1.475e-06,
|
367 |
-
"loss": 10.0019,
|
368 |
-
"step": 300
|
369 |
-
},
|
370 |
-
{
|
371 |
-
"epoch": 0.09,
|
372 |
-
"learning_rate": 1.5e-06,
|
373 |
-
"loss": 11.417,
|
374 |
-
"step": 305
|
375 |
-
},
|
376 |
-
{
|
377 |
-
"epoch": 0.09,
|
378 |
-
"learning_rate": 1.525e-06,
|
379 |
-
"loss": 8.4366,
|
380 |
-
"step": 310
|
381 |
-
},
|
382 |
-
{
|
383 |
-
"epoch": 0.09,
|
384 |
-
"learning_rate": 1.5500000000000002e-06,
|
385 |
-
"loss": 8.6453,
|
386 |
-
"step": 315
|
387 |
-
},
|
388 |
-
{
|
389 |
-
"epoch": 0.09,
|
390 |
-
"learning_rate": 1.5750000000000002e-06,
|
391 |
-
"loss": 7.6421,
|
392 |
-
"step": 320
|
393 |
-
},
|
394 |
-
{
|
395 |
-
"epoch": 0.09,
|
396 |
-
"learning_rate": 1.6000000000000001e-06,
|
397 |
-
"loss": 8.7056,
|
398 |
-
"step": 325
|
399 |
-
},
|
400 |
-
{
|
401 |
-
"epoch": 0.09,
|
402 |
-
"learning_rate": 1.6250000000000001e-06,
|
403 |
-
"loss": 7.4756,
|
404 |
-
"step": 330
|
405 |
-
},
|
406 |
-
{
|
407 |
-
"epoch": 0.09,
|
408 |
-
"learning_rate": 1.6500000000000003e-06,
|
409 |
-
"loss": 6.6272,
|
410 |
-
"step": 335
|
411 |
-
},
|
412 |
-
{
|
413 |
-
"epoch": 0.09,
|
414 |
-
"learning_rate": 1.6750000000000003e-06,
|
415 |
-
"loss": 6.9702,
|
416 |
-
"step": 340
|
417 |
-
},
|
418 |
-
{
|
419 |
-
"epoch": 0.1,
|
420 |
-
"learning_rate": 1.7000000000000002e-06,
|
421 |
-
"loss": 6.971,
|
422 |
-
"step": 345
|
423 |
-
},
|
424 |
-
{
|
425 |
-
"epoch": 0.1,
|
426 |
-
"learning_rate": 1.725e-06,
|
427 |
-
"loss": 6.4695,
|
428 |
-
"step": 350
|
429 |
-
},
|
430 |
-
{
|
431 |
-
"epoch": 0.1,
|
432 |
-
"learning_rate": 1.75e-06,
|
433 |
-
"loss": 9.6089,
|
434 |
-
"step": 355
|
435 |
-
},
|
436 |
-
{
|
437 |
-
"epoch": 0.1,
|
438 |
-
"learning_rate": 1.7750000000000002e-06,
|
439 |
-
"loss": 7.6838,
|
440 |
-
"step": 360
|
441 |
-
},
|
442 |
-
{
|
443 |
-
"epoch": 0.1,
|
444 |
-
"learning_rate": 1.8000000000000001e-06,
|
445 |
-
"loss": 6.6076,
|
446 |
-
"step": 365
|
447 |
-
},
|
448 |
-
{
|
449 |
-
"epoch": 0.1,
|
450 |
-
"learning_rate": 1.825e-06,
|
451 |
-
"loss": 8.1119,
|
452 |
-
"step": 370
|
453 |
-
},
|
454 |
-
{
|
455 |
-
"epoch": 0.1,
|
456 |
-
"learning_rate": 1.85e-06,
|
457 |
-
"loss": 8.6286,
|
458 |
-
"step": 375
|
459 |
-
},
|
460 |
-
{
|
461 |
-
"epoch": 0.11,
|
462 |
-
"learning_rate": 1.8750000000000003e-06,
|
463 |
-
"loss": 6.2768,
|
464 |
-
"step": 380
|
465 |
-
},
|
466 |
-
{
|
467 |
-
"epoch": 0.11,
|
468 |
-
"learning_rate": 1.9000000000000002e-06,
|
469 |
-
"loss": 6.9159,
|
470 |
-
"step": 385
|
471 |
-
},
|
472 |
-
{
|
473 |
-
"epoch": 0.11,
|
474 |
-
"learning_rate": 1.925e-06,
|
475 |
-
"loss": 7.0885,
|
476 |
-
"step": 390
|
477 |
-
},
|
478 |
-
{
|
479 |
-
"epoch": 0.11,
|
480 |
-
"learning_rate": 1.9500000000000004e-06,
|
481 |
-
"loss": 8.5395,
|
482 |
-
"step": 395
|
483 |
-
},
|
484 |
-
{
|
485 |
-
"epoch": 0.11,
|
486 |
-
"learning_rate": 1.975e-06,
|
487 |
-
"loss": 8.7365,
|
488 |
-
"step": 400
|
489 |
-
},
|
490 |
-
{
|
491 |
-
"epoch": 0.11,
|
492 |
-
"learning_rate": 2.0000000000000003e-06,
|
493 |
-
"loss": 8.7254,
|
494 |
-
"step": 405
|
495 |
-
},
|
496 |
-
{
|
497 |
-
"epoch": 0.11,
|
498 |
-
"learning_rate": 2.025e-06,
|
499 |
-
"loss": 6.4878,
|
500 |
-
"step": 410
|
501 |
-
},
|
502 |
-
{
|
503 |
-
"epoch": 0.12,
|
504 |
-
"learning_rate": 2.05e-06,
|
505 |
-
"loss": 10.4269,
|
506 |
-
"step": 415
|
507 |
-
},
|
508 |
-
{
|
509 |
-
"epoch": 0.12,
|
510 |
-
"learning_rate": 2.075e-06,
|
511 |
-
"loss": 6.2613,
|
512 |
-
"step": 420
|
513 |
-
},
|
514 |
-
{
|
515 |
-
"epoch": 0.12,
|
516 |
-
"learning_rate": 2.1000000000000002e-06,
|
517 |
-
"loss": 6.6001,
|
518 |
-
"step": 425
|
519 |
-
},
|
520 |
-
{
|
521 |
-
"epoch": 0.12,
|
522 |
-
"learning_rate": 2.125e-06,
|
523 |
-
"loss": 6.0585,
|
524 |
-
"step": 430
|
525 |
-
},
|
526 |
-
{
|
527 |
-
"epoch": 0.12,
|
528 |
-
"learning_rate": 2.15e-06,
|
529 |
-
"loss": 6.2137,
|
530 |
-
"step": 435
|
531 |
-
},
|
532 |
-
{
|
533 |
-
"epoch": 0.12,
|
534 |
-
"learning_rate": 2.1750000000000004e-06,
|
535 |
-
"loss": 8.026,
|
536 |
-
"step": 440
|
537 |
-
},
|
538 |
-
{
|
539 |
-
"epoch": 0.12,
|
540 |
-
"learning_rate": 2.2e-06,
|
541 |
-
"loss": 8.5144,
|
542 |
-
"step": 445
|
543 |
-
},
|
544 |
-
{
|
545 |
-
"epoch": 0.13,
|
546 |
-
"learning_rate": 2.2250000000000003e-06,
|
547 |
-
"loss": 6.092,
|
548 |
-
"step": 450
|
549 |
-
},
|
550 |
-
{
|
551 |
-
"epoch": 0.13,
|
552 |
-
"learning_rate": 2.25e-06,
|
553 |
-
"loss": 8.8018,
|
554 |
-
"step": 455
|
555 |
-
},
|
556 |
-
{
|
557 |
-
"epoch": 0.13,
|
558 |
-
"learning_rate": 2.2750000000000002e-06,
|
559 |
-
"loss": 6.2074,
|
560 |
-
"step": 460
|
561 |
-
},
|
562 |
-
{
|
563 |
-
"epoch": 0.13,
|
564 |
-
"learning_rate": 2.3000000000000004e-06,
|
565 |
-
"loss": 10.6086,
|
566 |
-
"step": 465
|
567 |
-
},
|
568 |
-
{
|
569 |
-
"epoch": 0.13,
|
570 |
-
"learning_rate": 2.325e-06,
|
571 |
-
"loss": 6.5399,
|
572 |
-
"step": 470
|
573 |
-
},
|
574 |
-
{
|
575 |
-
"epoch": 0.13,
|
576 |
-
"learning_rate": 2.35e-06,
|
577 |
-
"loss": 7.0942,
|
578 |
-
"step": 475
|
579 |
-
},
|
580 |
-
{
|
581 |
-
"epoch": 0.13,
|
582 |
-
"learning_rate": 2.375e-06,
|
583 |
-
"loss": 6.4347,
|
584 |
-
"step": 480
|
585 |
-
},
|
586 |
-
{
|
587 |
-
"epoch": 0.14,
|
588 |
-
"learning_rate": 2.4000000000000003e-06,
|
589 |
-
"loss": 7.1602,
|
590 |
-
"step": 485
|
591 |
-
},
|
592 |
-
{
|
593 |
-
"epoch": 0.14,
|
594 |
-
"learning_rate": 2.425e-06,
|
595 |
-
"loss": 5.4093,
|
596 |
-
"step": 490
|
597 |
-
},
|
598 |
-
{
|
599 |
-
"epoch": 0.14,
|
600 |
-
"learning_rate": 2.4500000000000003e-06,
|
601 |
-
"loss": 6.618,
|
602 |
-
"step": 495
|
603 |
-
},
|
604 |
-
{
|
605 |
-
"epoch": 0.14,
|
606 |
-
"learning_rate": 2.475e-06,
|
607 |
-
"loss": 8.3884,
|
608 |
-
"step": 500
|
609 |
-
},
|
610 |
-
{
|
611 |
-
"epoch": 0.14,
|
612 |
-
"eval_loss": 7.4285359382629395,
|
613 |
-
"eval_runtime": 109.7145,
|
614 |
-
"eval_samples_per_second": 10.883,
|
615 |
-
"eval_steps_per_second": 1.367,
|
616 |
-
"eval_wer": 1.0,
|
617 |
-
"step": 500
|
618 |
-
},
|
619 |
-
{
|
620 |
-
"epoch": 0.14,
|
621 |
-
"learning_rate": 2.5e-06,
|
622 |
-
"loss": 8.5122,
|
623 |
-
"step": 505
|
624 |
-
},
|
625 |
-
{
|
626 |
-
"epoch": 0.14,
|
627 |
-
"learning_rate": 2.5250000000000004e-06,
|
628 |
-
"loss": 9.0237,
|
629 |
-
"step": 510
|
630 |
-
},
|
631 |
-
{
|
632 |
-
"epoch": 0.14,
|
633 |
-
"learning_rate": 2.55e-06,
|
634 |
-
"loss": 6.343,
|
635 |
-
"step": 515
|
636 |
-
},
|
637 |
-
{
|
638 |
-
"epoch": 0.15,
|
639 |
-
"learning_rate": 2.5750000000000003e-06,
|
640 |
-
"loss": 6.3103,
|
641 |
-
"step": 520
|
642 |
-
},
|
643 |
-
{
|
644 |
-
"epoch": 0.15,
|
645 |
-
"learning_rate": 2.6e-06,
|
646 |
-
"loss": 7.6124,
|
647 |
-
"step": 525
|
648 |
-
},
|
649 |
-
{
|
650 |
-
"epoch": 0.15,
|
651 |
-
"learning_rate": 2.6250000000000003e-06,
|
652 |
-
"loss": 7.1252,
|
653 |
-
"step": 530
|
654 |
-
},
|
655 |
-
{
|
656 |
-
"epoch": 0.15,
|
657 |
-
"learning_rate": 2.6500000000000005e-06,
|
658 |
-
"loss": 7.0198,
|
659 |
-
"step": 535
|
660 |
-
},
|
661 |
-
{
|
662 |
-
"epoch": 0.15,
|
663 |
-
"learning_rate": 2.6750000000000002e-06,
|
664 |
-
"loss": 6.3734,
|
665 |
-
"step": 540
|
666 |
-
},
|
667 |
-
{
|
668 |
-
"epoch": 0.15,
|
669 |
-
"learning_rate": 2.7000000000000004e-06,
|
670 |
-
"loss": 4.7063,
|
671 |
-
"step": 545
|
672 |
-
},
|
673 |
-
{
|
674 |
-
"epoch": 0.15,
|
675 |
-
"learning_rate": 2.7250000000000006e-06,
|
676 |
-
"loss": 7.7081,
|
677 |
-
"step": 550
|
678 |
-
},
|
679 |
-
{
|
680 |
-
"epoch": 0.16,
|
681 |
-
"learning_rate": 2.7500000000000004e-06,
|
682 |
-
"loss": 5.9,
|
683 |
-
"step": 555
|
684 |
-
},
|
685 |
-
{
|
686 |
-
"epoch": 0.16,
|
687 |
-
"learning_rate": 2.7750000000000005e-06,
|
688 |
-
"loss": 5.634,
|
689 |
-
"step": 560
|
690 |
-
},
|
691 |
-
{
|
692 |
-
"epoch": 0.16,
|
693 |
-
"learning_rate": 2.8000000000000003e-06,
|
694 |
-
"loss": 8.6987,
|
695 |
-
"step": 565
|
696 |
-
},
|
697 |
-
{
|
698 |
-
"epoch": 0.16,
|
699 |
-
"learning_rate": 2.825e-06,
|
700 |
-
"loss": 6.5511,
|
701 |
-
"step": 570
|
702 |
-
},
|
703 |
-
{
|
704 |
-
"epoch": 0.16,
|
705 |
-
"learning_rate": 2.85e-06,
|
706 |
-
"loss": 7.6908,
|
707 |
-
"step": 575
|
708 |
-
},
|
709 |
-
{
|
710 |
-
"epoch": 0.16,
|
711 |
-
"learning_rate": 2.875e-06,
|
712 |
-
"loss": 7.4105,
|
713 |
-
"step": 580
|
714 |
-
},
|
715 |
-
{
|
716 |
-
"epoch": 0.16,
|
717 |
-
"learning_rate": 2.9e-06,
|
718 |
-
"loss": 5.802,
|
719 |
-
"step": 585
|
720 |
-
},
|
721 |
-
{
|
722 |
-
"epoch": 0.16,
|
723 |
-
"learning_rate": 2.925e-06,
|
724 |
-
"loss": 6.0835,
|
725 |
-
"step": 590
|
726 |
-
},
|
727 |
-
{
|
728 |
-
"epoch": 0.17,
|
729 |
-
"learning_rate": 2.95e-06,
|
730 |
-
"loss": 7.8744,
|
731 |
-
"step": 595
|
732 |
-
},
|
733 |
-
{
|
734 |
-
"epoch": 0.17,
|
735 |
-
"learning_rate": 2.9750000000000003e-06,
|
736 |
-
"loss": 6.1134,
|
737 |
-
"step": 600
|
738 |
-
},
|
739 |
-
{
|
740 |
-
"epoch": 0.17,
|
741 |
-
"learning_rate": 3e-06,
|
742 |
-
"loss": 6.6416,
|
743 |
-
"step": 605
|
744 |
-
},
|
745 |
-
{
|
746 |
-
"epoch": 0.17,
|
747 |
-
"learning_rate": 3.0250000000000003e-06,
|
748 |
-
"loss": 8.7177,
|
749 |
-
"step": 610
|
750 |
-
},
|
751 |
-
{
|
752 |
-
"epoch": 0.17,
|
753 |
-
"learning_rate": 3.05e-06,
|
754 |
-
"loss": 6.2383,
|
755 |
-
"step": 615
|
756 |
-
},
|
757 |
-
{
|
758 |
-
"epoch": 0.17,
|
759 |
-
"learning_rate": 3.075e-06,
|
760 |
-
"loss": 6.2413,
|
761 |
-
"step": 620
|
762 |
-
},
|
763 |
-
{
|
764 |
-
"epoch": 0.17,
|
765 |
-
"learning_rate": 3.1000000000000004e-06,
|
766 |
-
"loss": 5.4088,
|
767 |
-
"step": 625
|
768 |
-
},
|
769 |
-
{
|
770 |
-
"epoch": 0.18,
|
771 |
-
"learning_rate": 3.125e-06,
|
772 |
-
"loss": 4.3896,
|
773 |
-
"step": 630
|
774 |
-
},
|
775 |
-
{
|
776 |
-
"epoch": 0.18,
|
777 |
-
"learning_rate": 3.1500000000000003e-06,
|
778 |
-
"loss": 5.6138,
|
779 |
-
"step": 635
|
780 |
-
},
|
781 |
-
{
|
782 |
-
"epoch": 0.18,
|
783 |
-
"learning_rate": 3.175e-06,
|
784 |
-
"loss": 5.0078,
|
785 |
-
"step": 640
|
786 |
-
},
|
787 |
-
{
|
788 |
-
"epoch": 0.18,
|
789 |
-
"learning_rate": 3.2000000000000003e-06,
|
790 |
-
"loss": 6.9994,
|
791 |
-
"step": 645
|
792 |
-
},
|
793 |
-
{
|
794 |
-
"epoch": 0.18,
|
795 |
-
"learning_rate": 3.2250000000000005e-06,
|
796 |
-
"loss": 5.4662,
|
797 |
-
"step": 650
|
798 |
-
},
|
799 |
-
{
|
800 |
-
"epoch": 0.18,
|
801 |
-
"learning_rate": 3.2500000000000002e-06,
|
802 |
-
"loss": 5.144,
|
803 |
-
"step": 655
|
804 |
-
},
|
805 |
-
{
|
806 |
-
"epoch": 0.18,
|
807 |
-
"learning_rate": 3.2750000000000004e-06,
|
808 |
-
"loss": 5.4199,
|
809 |
-
"step": 660
|
810 |
-
},
|
811 |
-
{
|
812 |
-
"epoch": 0.19,
|
813 |
-
"learning_rate": 3.3000000000000006e-06,
|
814 |
-
"loss": 6.5677,
|
815 |
-
"step": 665
|
816 |
-
},
|
817 |
-
{
|
818 |
-
"epoch": 0.19,
|
819 |
-
"learning_rate": 3.3250000000000004e-06,
|
820 |
-
"loss": 6.168,
|
821 |
-
"step": 670
|
822 |
-
},
|
823 |
-
{
|
824 |
-
"epoch": 0.19,
|
825 |
-
"learning_rate": 3.3500000000000005e-06,
|
826 |
-
"loss": 6.3248,
|
827 |
-
"step": 675
|
828 |
-
},
|
829 |
-
{
|
830 |
-
"epoch": 0.19,
|
831 |
-
"learning_rate": 3.3750000000000003e-06,
|
832 |
-
"loss": 5.6306,
|
833 |
-
"step": 680
|
834 |
-
},
|
835 |
-
{
|
836 |
-
"epoch": 0.19,
|
837 |
-
"learning_rate": 3.4000000000000005e-06,
|
838 |
-
"loss": 4.4895,
|
839 |
-
"step": 685
|
840 |
-
},
|
841 |
-
{
|
842 |
-
"epoch": 0.19,
|
843 |
-
"learning_rate": 3.4250000000000007e-06,
|
844 |
-
"loss": 4.7549,
|
845 |
-
"step": 690
|
846 |
-
},
|
847 |
-
{
|
848 |
-
"epoch": 0.19,
|
849 |
-
"learning_rate": 3.45e-06,
|
850 |
-
"loss": 3.8521,
|
851 |
-
"step": 695
|
852 |
-
},
|
853 |
-
{
|
854 |
-
"epoch": 0.2,
|
855 |
-
"learning_rate": 3.475e-06,
|
856 |
-
"loss": 7.8289,
|
857 |
-
"step": 700
|
858 |
-
},
|
859 |
-
{
|
860 |
-
"epoch": 0.2,
|
861 |
-
"learning_rate": 3.5e-06,
|
862 |
-
"loss": 8.7039,
|
863 |
-
"step": 705
|
864 |
-
},
|
865 |
-
{
|
866 |
-
"epoch": 0.2,
|
867 |
-
"learning_rate": 3.525e-06,
|
868 |
-
"loss": 7.9723,
|
869 |
-
"step": 710
|
870 |
-
},
|
871 |
-
{
|
872 |
-
"epoch": 0.2,
|
873 |
-
"learning_rate": 3.5500000000000003e-06,
|
874 |
-
"loss": 7.4315,
|
875 |
-
"step": 715
|
876 |
-
},
|
877 |
-
{
|
878 |
-
"epoch": 0.2,
|
879 |
-
"learning_rate": 3.575e-06,
|
880 |
-
"loss": 5.6061,
|
881 |
-
"step": 720
|
882 |
-
},
|
883 |
-
{
|
884 |
-
"epoch": 0.2,
|
885 |
-
"learning_rate": 3.6000000000000003e-06,
|
886 |
-
"loss": 5.4848,
|
887 |
-
"step": 725
|
888 |
-
},
|
889 |
-
{
|
890 |
-
"epoch": 0.2,
|
891 |
-
"learning_rate": 3.625e-06,
|
892 |
-
"loss": 4.1425,
|
893 |
-
"step": 730
|
894 |
-
},
|
895 |
-
{
|
896 |
-
"epoch": 0.21,
|
897 |
-
"learning_rate": 3.65e-06,
|
898 |
-
"loss": 5.5031,
|
899 |
-
"step": 735
|
900 |
-
},
|
901 |
-
{
|
902 |
-
"epoch": 0.21,
|
903 |
-
"learning_rate": 3.6750000000000004e-06,
|
904 |
-
"loss": 4.4556,
|
905 |
-
"step": 740
|
906 |
-
},
|
907 |
-
{
|
908 |
-
"epoch": 0.21,
|
909 |
-
"learning_rate": 3.7e-06,
|
910 |
-
"loss": 5.9729,
|
911 |
-
"step": 745
|
912 |
-
},
|
913 |
-
{
|
914 |
-
"epoch": 0.21,
|
915 |
-
"learning_rate": 3.7250000000000003e-06,
|
916 |
-
"loss": 4.5424,
|
917 |
-
"step": 750
|
918 |
-
},
|
919 |
-
{
|
920 |
-
"epoch": 0.21,
|
921 |
-
"learning_rate": 3.7500000000000005e-06,
|
922 |
-
"loss": 7.1968,
|
923 |
-
"step": 755
|
924 |
-
},
|
925 |
-
{
|
926 |
-
"epoch": 0.21,
|
927 |
-
"learning_rate": 3.7750000000000003e-06,
|
928 |
-
"loss": 6.0854,
|
929 |
-
"step": 760
|
930 |
-
},
|
931 |
-
{
|
932 |
-
"epoch": 0.21,
|
933 |
-
"learning_rate": 3.8000000000000005e-06,
|
934 |
-
"loss": 5.1684,
|
935 |
-
"step": 765
|
936 |
-
},
|
937 |
-
{
|
938 |
-
"epoch": 0.22,
|
939 |
-
"learning_rate": 3.825000000000001e-06,
|
940 |
-
"loss": 4.8008,
|
941 |
-
"step": 770
|
942 |
-
},
|
943 |
-
{
|
944 |
-
"epoch": 0.22,
|
945 |
-
"learning_rate": 3.85e-06,
|
946 |
-
"loss": 5.2927,
|
947 |
-
"step": 775
|
948 |
-
},
|
949 |
-
{
|
950 |
-
"epoch": 0.22,
|
951 |
-
"learning_rate": 3.875e-06,
|
952 |
-
"loss": 4.1035,
|
953 |
-
"step": 780
|
954 |
-
},
|
955 |
-
{
|
956 |
-
"epoch": 0.22,
|
957 |
-
"learning_rate": 3.900000000000001e-06,
|
958 |
-
"loss": 4.083,
|
959 |
-
"step": 785
|
960 |
-
},
|
961 |
-
{
|
962 |
-
"epoch": 0.22,
|
963 |
-
"learning_rate": 3.9250000000000005e-06,
|
964 |
-
"loss": 4.2797,
|
965 |
-
"step": 790
|
966 |
-
},
|
967 |
-
{
|
968 |
-
"epoch": 0.22,
|
969 |
-
"learning_rate": 3.95e-06,
|
970 |
-
"loss": 6.4263,
|
971 |
-
"step": 795
|
972 |
-
},
|
973 |
-
{
|
974 |
-
"epoch": 0.22,
|
975 |
-
"learning_rate": 3.975000000000001e-06,
|
976 |
-
"loss": 6.5447,
|
977 |
-
"step": 800
|
978 |
-
},
|
979 |
-
{
|
980 |
-
"epoch": 0.22,
|
981 |
-
"learning_rate": 4.000000000000001e-06,
|
982 |
-
"loss": 5.9181,
|
983 |
-
"step": 805
|
984 |
-
},
|
985 |
-
{
|
986 |
-
"epoch": 0.23,
|
987 |
-
"learning_rate": 4.0250000000000004e-06,
|
988 |
-
"loss": 6.3199,
|
989 |
-
"step": 810
|
990 |
-
},
|
991 |
-
{
|
992 |
-
"epoch": 0.23,
|
993 |
-
"learning_rate": 4.05e-06,
|
994 |
-
"loss": 7.4045,
|
995 |
-
"step": 815
|
996 |
-
},
|
997 |
-
{
|
998 |
-
"epoch": 0.23,
|
999 |
-
"learning_rate": 4.075e-06,
|
1000 |
-
"loss": 4.7224,
|
1001 |
-
"step": 820
|
1002 |
-
},
|
1003 |
-
{
|
1004 |
-
"epoch": 0.23,
|
1005 |
-
"learning_rate": 4.1e-06,
|
1006 |
-
"loss": 4.3196,
|
1007 |
-
"step": 825
|
1008 |
-
},
|
1009 |
-
{
|
1010 |
-
"epoch": 0.23,
|
1011 |
-
"learning_rate": 4.125e-06,
|
1012 |
-
"loss": 6.7817,
|
1013 |
-
"step": 830
|
1014 |
-
},
|
1015 |
-
{
|
1016 |
-
"epoch": 0.23,
|
1017 |
-
"learning_rate": 4.15e-06,
|
1018 |
-
"loss": 3.9509,
|
1019 |
-
"step": 835
|
1020 |
-
},
|
1021 |
-
{
|
1022 |
-
"epoch": 0.23,
|
1023 |
-
"learning_rate": 4.175e-06,
|
1024 |
-
"loss": 5.2207,
|
1025 |
-
"step": 840
|
1026 |
-
},
|
1027 |
-
{
|
1028 |
-
"epoch": 0.24,
|
1029 |
-
"learning_rate": 4.2000000000000004e-06,
|
1030 |
-
"loss": 4.5517,
|
1031 |
-
"step": 845
|
1032 |
-
},
|
1033 |
-
{
|
1034 |
-
"epoch": 0.24,
|
1035 |
-
"learning_rate": 4.225e-06,
|
1036 |
-
"loss": 5.7441,
|
1037 |
-
"step": 850
|
1038 |
-
},
|
1039 |
-
{
|
1040 |
-
"epoch": 0.24,
|
1041 |
-
"learning_rate": 4.25e-06,
|
1042 |
-
"loss": 6.9507,
|
1043 |
-
"step": 855
|
1044 |
-
},
|
1045 |
-
{
|
1046 |
-
"epoch": 0.24,
|
1047 |
-
"learning_rate": 4.2750000000000006e-06,
|
1048 |
-
"loss": 4.8579,
|
1049 |
-
"step": 860
|
1050 |
-
},
|
1051 |
-
{
|
1052 |
-
"epoch": 0.24,
|
1053 |
-
"learning_rate": 4.3e-06,
|
1054 |
-
"loss": 5.6729,
|
1055 |
-
"step": 865
|
1056 |
-
},
|
1057 |
-
{
|
1058 |
-
"epoch": 0.24,
|
1059 |
-
"learning_rate": 4.325e-06,
|
1060 |
-
"loss": 5.7697,
|
1061 |
-
"step": 870
|
1062 |
-
},
|
1063 |
-
{
|
1064 |
-
"epoch": 0.24,
|
1065 |
-
"learning_rate": 4.350000000000001e-06,
|
1066 |
-
"loss": 4.7595,
|
1067 |
-
"step": 875
|
1068 |
-
},
|
1069 |
-
{
|
1070 |
-
"epoch": 0.25,
|
1071 |
-
"learning_rate": 4.3750000000000005e-06,
|
1072 |
-
"loss": 4.6507,
|
1073 |
-
"step": 880
|
1074 |
-
},
|
1075 |
-
{
|
1076 |
-
"epoch": 0.25,
|
1077 |
-
"learning_rate": 4.4e-06,
|
1078 |
-
"loss": 5.0327,
|
1079 |
-
"step": 885
|
1080 |
-
},
|
1081 |
-
{
|
1082 |
-
"epoch": 0.25,
|
1083 |
-
"learning_rate": 4.425e-06,
|
1084 |
-
"loss": 4.3302,
|
1085 |
-
"step": 890
|
1086 |
-
},
|
1087 |
-
{
|
1088 |
-
"epoch": 0.25,
|
1089 |
-
"learning_rate": 4.450000000000001e-06,
|
1090 |
-
"loss": 3.2291,
|
1091 |
-
"step": 895
|
1092 |
-
},
|
1093 |
-
{
|
1094 |
-
"epoch": 0.25,
|
1095 |
-
"learning_rate": 4.475e-06,
|
1096 |
-
"loss": 3.5748,
|
1097 |
-
"step": 900
|
1098 |
-
},
|
1099 |
-
{
|
1100 |
-
"epoch": 0.25,
|
1101 |
-
"learning_rate": 4.5e-06,
|
1102 |
-
"loss": 2.9844,
|
1103 |
-
"step": 905
|
1104 |
-
},
|
1105 |
-
{
|
1106 |
-
"epoch": 0.25,
|
1107 |
-
"learning_rate": 4.525000000000001e-06,
|
1108 |
-
"loss": 3.2969,
|
1109 |
-
"step": 910
|
1110 |
-
},
|
1111 |
-
{
|
1112 |
-
"epoch": 0.26,
|
1113 |
-
"learning_rate": 4.5500000000000005e-06,
|
1114 |
-
"loss": 3.2062,
|
1115 |
-
"step": 915
|
1116 |
-
},
|
1117 |
-
{
|
1118 |
-
"epoch": 0.26,
|
1119 |
-
"learning_rate": 4.575e-06,
|
1120 |
-
"loss": 3.1091,
|
1121 |
-
"step": 920
|
1122 |
-
},
|
1123 |
-
{
|
1124 |
-
"epoch": 0.26,
|
1125 |
-
"learning_rate": 4.600000000000001e-06,
|
1126 |
-
"loss": 3.288,
|
1127 |
-
"step": 925
|
1128 |
-
},
|
1129 |
-
{
|
1130 |
-
"epoch": 0.26,
|
1131 |
-
"learning_rate": 4.625000000000001e-06,
|
1132 |
-
"loss": 2.8966,
|
1133 |
-
"step": 930
|
1134 |
-
},
|
1135 |
-
{
|
1136 |
-
"epoch": 0.26,
|
1137 |
-
"learning_rate": 4.65e-06,
|
1138 |
-
"loss": 2.9091,
|
1139 |
-
"step": 935
|
1140 |
-
},
|
1141 |
-
{
|
1142 |
-
"epoch": 0.26,
|
1143 |
-
"learning_rate": 4.675000000000001e-06,
|
1144 |
-
"loss": 3.212,
|
1145 |
-
"step": 940
|
1146 |
-
},
|
1147 |
-
{
|
1148 |
-
"epoch": 0.26,
|
1149 |
-
"learning_rate": 4.7e-06,
|
1150 |
-
"loss": 2.8426,
|
1151 |
-
"step": 945
|
1152 |
-
},
|
1153 |
-
{
|
1154 |
-
"epoch": 0.27,
|
1155 |
-
"learning_rate": 4.7250000000000005e-06,
|
1156 |
-
"loss": 3.1961,
|
1157 |
-
"step": 950
|
1158 |
-
},
|
1159 |
-
{
|
1160 |
-
"epoch": 0.27,
|
1161 |
-
"learning_rate": 4.75e-06,
|
1162 |
-
"loss": 2.9483,
|
1163 |
-
"step": 955
|
1164 |
-
},
|
1165 |
-
{
|
1166 |
-
"epoch": 0.27,
|
1167 |
-
"learning_rate": 4.775e-06,
|
1168 |
-
"loss": 2.9817,
|
1169 |
-
"step": 960
|
1170 |
-
},
|
1171 |
-
{
|
1172 |
-
"epoch": 0.27,
|
1173 |
-
"learning_rate": 4.800000000000001e-06,
|
1174 |
-
"loss": 3.0139,
|
1175 |
-
"step": 965
|
1176 |
-
},
|
1177 |
-
{
|
1178 |
-
"epoch": 0.27,
|
1179 |
-
"learning_rate": 4.825e-06,
|
1180 |
-
"loss": 2.8438,
|
1181 |
-
"step": 970
|
1182 |
-
},
|
1183 |
-
{
|
1184 |
-
"epoch": 0.27,
|
1185 |
-
"learning_rate": 4.85e-06,
|
1186 |
-
"loss": 3.008,
|
1187 |
-
"step": 975
|
1188 |
-
},
|
1189 |
-
{
|
1190 |
-
"epoch": 0.27,
|
1191 |
-
"learning_rate": 4.875e-06,
|
1192 |
-
"loss": 2.9319,
|
1193 |
-
"step": 980
|
1194 |
-
},
|
1195 |
-
{
|
1196 |
-
"epoch": 0.28,
|
1197 |
-
"learning_rate": 4.9000000000000005e-06,
|
1198 |
-
"loss": 2.8704,
|
1199 |
-
"step": 985
|
1200 |
-
},
|
1201 |
-
{
|
1202 |
-
"epoch": 0.28,
|
1203 |
-
"learning_rate": 4.925e-06,
|
1204 |
-
"loss": 3.0068,
|
1205 |
-
"step": 990
|
1206 |
-
},
|
1207 |
-
{
|
1208 |
-
"epoch": 0.28,
|
1209 |
-
"learning_rate": 4.95e-06,
|
1210 |
-
"loss": 2.8494,
|
1211 |
-
"step": 995
|
1212 |
-
},
|
1213 |
-
{
|
1214 |
-
"epoch": 0.28,
|
1215 |
-
"learning_rate": 4.975000000000001e-06,
|
1216 |
-
"loss": 2.8652,
|
1217 |
-
"step": 1000
|
1218 |
-
},
|
1219 |
-
{
|
1220 |
-
"epoch": 0.28,
|
1221 |
-
"eval_loss": 2.881692886352539,
|
1222 |
-
"eval_runtime": 141.6494,
|
1223 |
-
"eval_samples_per_second": 8.429,
|
1224 |
-
"eval_steps_per_second": 1.059,
|
1225 |
-
"eval_wer": 1.0,
|
1226 |
-
"step": 1000
|
1227 |
-
},
|
1228 |
-
{
|
1229 |
-
"epoch": 0.28,
|
1230 |
-
"learning_rate": 5e-06,
|
1231 |
-
"loss": 2.9299,
|
1232 |
-
"step": 1005
|
1233 |
-
},
|
1234 |
-
{
|
1235 |
-
"epoch": 0.28,
|
1236 |
-
"learning_rate": 4.998122559327126e-06,
|
1237 |
-
"loss": 2.855,
|
1238 |
-
"step": 1010
|
1239 |
-
},
|
1240 |
-
{
|
1241 |
-
"epoch": 0.28,
|
1242 |
-
"learning_rate": 4.996245118654251e-06,
|
1243 |
-
"loss": 2.8683,
|
1244 |
-
"step": 1015
|
1245 |
-
},
|
1246 |
-
{
|
1247 |
-
"epoch": 0.28,
|
1248 |
-
"learning_rate": 4.994367677981376e-06,
|
1249 |
-
"loss": 2.8563,
|
1250 |
-
"step": 1020
|
1251 |
-
},
|
1252 |
-
{
|
1253 |
-
"epoch": 0.29,
|
1254 |
-
"learning_rate": 4.992490237308501e-06,
|
1255 |
-
"loss": 2.9462,
|
1256 |
-
"step": 1025
|
1257 |
-
},
|
1258 |
-
{
|
1259 |
-
"epoch": 0.29,
|
1260 |
-
"learning_rate": 4.990612796635627e-06,
|
1261 |
-
"loss": 2.8713,
|
1262 |
-
"step": 1030
|
1263 |
-
},
|
1264 |
-
{
|
1265 |
-
"epoch": 0.29,
|
1266 |
-
"learning_rate": 4.988735355962752e-06,
|
1267 |
-
"loss": 2.9464,
|
1268 |
-
"step": 1035
|
1269 |
-
},
|
1270 |
-
{
|
1271 |
-
"epoch": 0.29,
|
1272 |
-
"learning_rate": 4.986857915289877e-06,
|
1273 |
-
"loss": 2.8587,
|
1274 |
-
"step": 1040
|
1275 |
-
},
|
1276 |
-
{
|
1277 |
-
"epoch": 0.29,
|
1278 |
-
"learning_rate": 4.984980474617002e-06,
|
1279 |
-
"loss": 2.9491,
|
1280 |
-
"step": 1045
|
1281 |
-
},
|
1282 |
-
{
|
1283 |
-
"epoch": 0.29,
|
1284 |
-
"learning_rate": 4.9831030339441276e-06,
|
1285 |
-
"loss": 2.8847,
|
1286 |
-
"step": 1050
|
1287 |
-
},
|
1288 |
-
{
|
1289 |
-
"epoch": 0.29,
|
1290 |
-
"learning_rate": 4.981225593271253e-06,
|
1291 |
-
"loss": 2.9443,
|
1292 |
-
"step": 1055
|
1293 |
-
},
|
1294 |
-
{
|
1295 |
-
"epoch": 0.3,
|
1296 |
-
"learning_rate": 4.979348152598378e-06,
|
1297 |
-
"loss": 2.9,
|
1298 |
-
"step": 1060
|
1299 |
-
},
|
1300 |
-
{
|
1301 |
-
"epoch": 0.3,
|
1302 |
-
"learning_rate": 4.977470711925503e-06,
|
1303 |
-
"loss": 2.8448,
|
1304 |
-
"step": 1065
|
1305 |
-
},
|
1306 |
-
{
|
1307 |
-
"epoch": 0.3,
|
1308 |
-
"learning_rate": 4.9755932712526285e-06,
|
1309 |
-
"loss": 2.862,
|
1310 |
-
"step": 1070
|
1311 |
-
},
|
1312 |
-
{
|
1313 |
-
"epoch": 0.3,
|
1314 |
-
"learning_rate": 4.973715830579754e-06,
|
1315 |
-
"loss": 2.8782,
|
1316 |
-
"step": 1075
|
1317 |
-
},
|
1318 |
-
{
|
1319 |
-
"epoch": 0.3,
|
1320 |
-
"learning_rate": 4.97183838990688e-06,
|
1321 |
-
"loss": 2.866,
|
1322 |
-
"step": 1080
|
1323 |
-
},
|
1324 |
-
{
|
1325 |
-
"epoch": 0.3,
|
1326 |
-
"learning_rate": 4.969960949234004e-06,
|
1327 |
-
"loss": 2.9187,
|
1328 |
-
"step": 1085
|
1329 |
-
},
|
1330 |
-
{
|
1331 |
-
"epoch": 0.3,
|
1332 |
-
"learning_rate": 4.96808350856113e-06,
|
1333 |
-
"loss": 2.8627,
|
1334 |
-
"step": 1090
|
1335 |
-
},
|
1336 |
-
{
|
1337 |
-
"epoch": 0.31,
|
1338 |
-
"learning_rate": 4.966206067888255e-06,
|
1339 |
-
"loss": 2.8636,
|
1340 |
-
"step": 1095
|
1341 |
-
},
|
1342 |
-
{
|
1343 |
-
"epoch": 0.31,
|
1344 |
-
"learning_rate": 4.964328627215381e-06,
|
1345 |
-
"loss": 2.9201,
|
1346 |
-
"step": 1100
|
1347 |
-
},
|
1348 |
-
{
|
1349 |
-
"epoch": 0.31,
|
1350 |
-
"learning_rate": 4.962451186542506e-06,
|
1351 |
-
"loss": 2.9078,
|
1352 |
-
"step": 1105
|
1353 |
-
},
|
1354 |
-
{
|
1355 |
-
"epoch": 0.31,
|
1356 |
-
"learning_rate": 4.9605737458696305e-06,
|
1357 |
-
"loss": 2.8853,
|
1358 |
-
"step": 1110
|
1359 |
-
},
|
1360 |
-
{
|
1361 |
-
"epoch": 0.31,
|
1362 |
-
"learning_rate": 4.9586963051967565e-06,
|
1363 |
-
"loss": 2.8612,
|
1364 |
-
"step": 1115
|
1365 |
-
},
|
1366 |
-
{
|
1367 |
-
"epoch": 0.31,
|
1368 |
-
"learning_rate": 4.956818864523881e-06,
|
1369 |
-
"loss": 2.8607,
|
1370 |
-
"step": 1120
|
1371 |
-
},
|
1372 |
-
{
|
1373 |
-
"epoch": 0.31,
|
1374 |
-
"learning_rate": 4.954941423851007e-06,
|
1375 |
-
"loss": 2.845,
|
1376 |
-
"step": 1125
|
1377 |
-
},
|
1378 |
-
{
|
1379 |
-
"epoch": 0.32,
|
1380 |
-
"learning_rate": 4.953063983178132e-06,
|
1381 |
-
"loss": 2.8778,
|
1382 |
-
"step": 1130
|
1383 |
-
},
|
1384 |
-
{
|
1385 |
-
"epoch": 0.32,
|
1386 |
-
"learning_rate": 4.9511865425052575e-06,
|
1387 |
-
"loss": 3.1876,
|
1388 |
-
"step": 1135
|
1389 |
-
},
|
1390 |
-
{
|
1391 |
-
"epoch": 0.32,
|
1392 |
-
"learning_rate": 4.949309101832383e-06,
|
1393 |
-
"loss": 2.8603,
|
1394 |
-
"step": 1140
|
1395 |
-
},
|
1396 |
-
{
|
1397 |
-
"epoch": 0.32,
|
1398 |
-
"learning_rate": 4.947431661159508e-06,
|
1399 |
-
"loss": 2.8403,
|
1400 |
-
"step": 1145
|
1401 |
-
},
|
1402 |
-
{
|
1403 |
-
"epoch": 0.32,
|
1404 |
-
"learning_rate": 4.945554220486633e-06,
|
1405 |
-
"loss": 2.8791,
|
1406 |
-
"step": 1150
|
1407 |
-
},
|
1408 |
-
{
|
1409 |
-
"epoch": 0.32,
|
1410 |
-
"learning_rate": 4.9436767798137585e-06,
|
1411 |
-
"loss": 2.8604,
|
1412 |
-
"step": 1155
|
1413 |
-
},
|
1414 |
-
{
|
1415 |
-
"epoch": 0.32,
|
1416 |
-
"learning_rate": 4.941799339140884e-06,
|
1417 |
-
"loss": 2.8595,
|
1418 |
-
"step": 1160
|
1419 |
-
},
|
1420 |
-
{
|
1421 |
-
"epoch": 0.33,
|
1422 |
-
"learning_rate": 4.939921898468009e-06,
|
1423 |
-
"loss": 2.9011,
|
1424 |
-
"step": 1165
|
1425 |
-
},
|
1426 |
-
{
|
1427 |
-
"epoch": 0.33,
|
1428 |
-
"learning_rate": 4.938044457795134e-06,
|
1429 |
-
"loss": 2.86,
|
1430 |
-
"step": 1170
|
1431 |
-
},
|
1432 |
-
{
|
1433 |
-
"epoch": 0.33,
|
1434 |
-
"learning_rate": 4.936167017122259e-06,
|
1435 |
-
"loss": 2.8583,
|
1436 |
-
"step": 1175
|
1437 |
-
},
|
1438 |
-
{
|
1439 |
-
"epoch": 0.33,
|
1440 |
-
"learning_rate": 4.934289576449385e-06,
|
1441 |
-
"loss": 2.9274,
|
1442 |
-
"step": 1180
|
1443 |
-
},
|
1444 |
-
{
|
1445 |
-
"epoch": 0.33,
|
1446 |
-
"learning_rate": 4.93241213577651e-06,
|
1447 |
-
"loss": 2.9472,
|
1448 |
-
"step": 1185
|
1449 |
-
},
|
1450 |
-
{
|
1451 |
-
"epoch": 0.33,
|
1452 |
-
"learning_rate": 4.930534695103635e-06,
|
1453 |
-
"loss": 3.4067,
|
1454 |
-
"step": 1190
|
1455 |
-
},
|
1456 |
-
{
|
1457 |
-
"epoch": 0.33,
|
1458 |
-
"learning_rate": 4.92865725443076e-06,
|
1459 |
-
"loss": 2.9363,
|
1460 |
-
"step": 1195
|
1461 |
-
},
|
1462 |
-
{
|
1463 |
-
"epoch": 0.34,
|
1464 |
-
"learning_rate": 4.926779813757886e-06,
|
1465 |
-
"loss": 2.8608,
|
1466 |
-
"step": 1200
|
1467 |
-
},
|
1468 |
-
{
|
1469 |
-
"epoch": 0.34,
|
1470 |
-
"learning_rate": 4.924902373085011e-06,
|
1471 |
-
"loss": 3.356,
|
1472 |
-
"step": 1205
|
1473 |
-
},
|
1474 |
-
{
|
1475 |
-
"epoch": 0.34,
|
1476 |
-
"learning_rate": 4.923024932412136e-06,
|
1477 |
-
"loss": 2.8678,
|
1478 |
-
"step": 1210
|
1479 |
-
},
|
1480 |
-
{
|
1481 |
-
"epoch": 0.34,
|
1482 |
-
"learning_rate": 4.921147491739261e-06,
|
1483 |
-
"loss": 2.8664,
|
1484 |
-
"step": 1215
|
1485 |
-
},
|
1486 |
-
{
|
1487 |
-
"epoch": 0.34,
|
1488 |
-
"learning_rate": 4.919270051066387e-06,
|
1489 |
-
"loss": 2.8897,
|
1490 |
-
"step": 1220
|
1491 |
-
},
|
1492 |
-
{
|
1493 |
-
"epoch": 0.34,
|
1494 |
-
"learning_rate": 4.917392610393512e-06,
|
1495 |
-
"loss": 2.8783,
|
1496 |
-
"step": 1225
|
1497 |
-
},
|
1498 |
-
{
|
1499 |
-
"epoch": 0.34,
|
1500 |
-
"learning_rate": 4.915515169720637e-06,
|
1501 |
-
"loss": 2.8886,
|
1502 |
-
"step": 1230
|
1503 |
-
},
|
1504 |
-
{
|
1505 |
-
"epoch": 0.35,
|
1506 |
-
"learning_rate": 4.913637729047762e-06,
|
1507 |
-
"loss": 2.8534,
|
1508 |
-
"step": 1235
|
1509 |
-
},
|
1510 |
-
{
|
1511 |
-
"epoch": 0.35,
|
1512 |
-
"learning_rate": 4.9117602883748875e-06,
|
1513 |
-
"loss": 2.8322,
|
1514 |
-
"step": 1240
|
1515 |
-
},
|
1516 |
-
{
|
1517 |
-
"epoch": 0.35,
|
1518 |
-
"learning_rate": 4.909882847702014e-06,
|
1519 |
-
"loss": 2.8148,
|
1520 |
-
"step": 1245
|
1521 |
-
},
|
1522 |
-
{
|
1523 |
-
"epoch": 0.35,
|
1524 |
-
"learning_rate": 4.908005407029138e-06,
|
1525 |
-
"loss": 2.832,
|
1526 |
-
"step": 1250
|
1527 |
-
},
|
1528 |
-
{
|
1529 |
-
"epoch": 0.35,
|
1530 |
-
"learning_rate": 4.906127966356263e-06,
|
1531 |
-
"loss": 2.8783,
|
1532 |
-
"step": 1255
|
1533 |
-
},
|
1534 |
-
{
|
1535 |
-
"epoch": 0.35,
|
1536 |
-
"learning_rate": 4.9042505256833885e-06,
|
1537 |
-
"loss": 2.8255,
|
1538 |
-
"step": 1260
|
1539 |
-
},
|
1540 |
-
{
|
1541 |
-
"epoch": 0.35,
|
1542 |
-
"learning_rate": 4.902373085010514e-06,
|
1543 |
-
"loss": 2.8893,
|
1544 |
-
"step": 1265
|
1545 |
-
},
|
1546 |
-
{
|
1547 |
-
"epoch": 0.35,
|
1548 |
-
"learning_rate": 4.90049564433764e-06,
|
1549 |
-
"loss": 2.9132,
|
1550 |
-
"step": 1270
|
1551 |
-
},
|
1552 |
-
{
|
1553 |
-
"epoch": 0.36,
|
1554 |
-
"learning_rate": 4.898618203664764e-06,
|
1555 |
-
"loss": 2.8366,
|
1556 |
-
"step": 1275
|
1557 |
-
},
|
1558 |
-
{
|
1559 |
-
"epoch": 0.36,
|
1560 |
-
"learning_rate": 4.89674076299189e-06,
|
1561 |
-
"loss": 2.8717,
|
1562 |
-
"step": 1280
|
1563 |
-
},
|
1564 |
-
{
|
1565 |
-
"epoch": 0.36,
|
1566 |
-
"learning_rate": 4.894863322319015e-06,
|
1567 |
-
"loss": 3.0332,
|
1568 |
-
"step": 1285
|
1569 |
-
},
|
1570 |
-
{
|
1571 |
-
"epoch": 0.36,
|
1572 |
-
"learning_rate": 4.892985881646141e-06,
|
1573 |
-
"loss": 2.8806,
|
1574 |
-
"step": 1290
|
1575 |
-
},
|
1576 |
-
{
|
1577 |
-
"epoch": 0.36,
|
1578 |
-
"learning_rate": 4.891108440973266e-06,
|
1579 |
-
"loss": 2.8377,
|
1580 |
-
"step": 1295
|
1581 |
-
},
|
1582 |
-
{
|
1583 |
-
"epoch": 0.36,
|
1584 |
-
"learning_rate": 4.88923100030039e-06,
|
1585 |
-
"loss": 2.8136,
|
1586 |
-
"step": 1300
|
1587 |
-
},
|
1588 |
-
{
|
1589 |
-
"epoch": 0.36,
|
1590 |
-
"learning_rate": 4.8873535596275165e-06,
|
1591 |
-
"loss": 2.9869,
|
1592 |
-
"step": 1305
|
1593 |
-
},
|
1594 |
-
{
|
1595 |
-
"epoch": 0.37,
|
1596 |
-
"learning_rate": 4.885476118954641e-06,
|
1597 |
-
"loss": 2.8812,
|
1598 |
-
"step": 1310
|
1599 |
-
},
|
1600 |
-
{
|
1601 |
-
"epoch": 0.37,
|
1602 |
-
"learning_rate": 4.883598678281767e-06,
|
1603 |
-
"loss": 2.83,
|
1604 |
-
"step": 1315
|
1605 |
-
},
|
1606 |
-
{
|
1607 |
-
"epoch": 0.37,
|
1608 |
-
"learning_rate": 4.881721237608892e-06,
|
1609 |
-
"loss": 2.8324,
|
1610 |
-
"step": 1320
|
1611 |
-
},
|
1612 |
-
{
|
1613 |
-
"epoch": 0.37,
|
1614 |
-
"learning_rate": 4.8798437969360175e-06,
|
1615 |
-
"loss": 2.829,
|
1616 |
-
"step": 1325
|
1617 |
-
},
|
1618 |
-
{
|
1619 |
-
"epoch": 0.37,
|
1620 |
-
"learning_rate": 4.877966356263143e-06,
|
1621 |
-
"loss": 2.834,
|
1622 |
-
"step": 1330
|
1623 |
-
},
|
1624 |
-
{
|
1625 |
-
"epoch": 0.37,
|
1626 |
-
"learning_rate": 4.876088915590268e-06,
|
1627 |
-
"loss": 3.1544,
|
1628 |
-
"step": 1335
|
1629 |
-
},
|
1630 |
-
{
|
1631 |
-
"epoch": 0.37,
|
1632 |
-
"learning_rate": 4.874211474917393e-06,
|
1633 |
-
"loss": 2.8126,
|
1634 |
-
"step": 1340
|
1635 |
-
},
|
1636 |
-
{
|
1637 |
-
"epoch": 0.38,
|
1638 |
-
"learning_rate": 4.8723340342445184e-06,
|
1639 |
-
"loss": 2.8915,
|
1640 |
-
"step": 1345
|
1641 |
-
},
|
1642 |
-
{
|
1643 |
-
"epoch": 0.38,
|
1644 |
-
"learning_rate": 4.870456593571644e-06,
|
1645 |
-
"loss": 3.0785,
|
1646 |
-
"step": 1350
|
1647 |
-
},
|
1648 |
-
{
|
1649 |
-
"epoch": 0.38,
|
1650 |
-
"learning_rate": 4.868579152898769e-06,
|
1651 |
-
"loss": 2.8628,
|
1652 |
-
"step": 1355
|
1653 |
-
},
|
1654 |
-
{
|
1655 |
-
"epoch": 0.38,
|
1656 |
-
"learning_rate": 4.866701712225894e-06,
|
1657 |
-
"loss": 2.8263,
|
1658 |
-
"step": 1360
|
1659 |
-
},
|
1660 |
-
{
|
1661 |
-
"epoch": 0.38,
|
1662 |
-
"learning_rate": 4.864824271553019e-06,
|
1663 |
-
"loss": 2.8055,
|
1664 |
-
"step": 1365
|
1665 |
-
},
|
1666 |
-
{
|
1667 |
-
"epoch": 0.38,
|
1668 |
-
"learning_rate": 4.862946830880145e-06,
|
1669 |
-
"loss": 2.837,
|
1670 |
-
"step": 1370
|
1671 |
-
},
|
1672 |
-
{
|
1673 |
-
"epoch": 0.38,
|
1674 |
-
"learning_rate": 4.86106939020727e-06,
|
1675 |
-
"loss": 2.8109,
|
1676 |
-
"step": 1375
|
1677 |
-
},
|
1678 |
-
{
|
1679 |
-
"epoch": 0.39,
|
1680 |
-
"learning_rate": 4.859191949534395e-06,
|
1681 |
-
"loss": 2.8313,
|
1682 |
-
"step": 1380
|
1683 |
-
},
|
1684 |
-
{
|
1685 |
-
"epoch": 0.39,
|
1686 |
-
"learning_rate": 4.85731450886152e-06,
|
1687 |
-
"loss": 2.8085,
|
1688 |
-
"step": 1385
|
1689 |
-
},
|
1690 |
-
{
|
1691 |
-
"epoch": 0.39,
|
1692 |
-
"learning_rate": 4.855437068188646e-06,
|
1693 |
-
"loss": 2.9006,
|
1694 |
-
"step": 1390
|
1695 |
-
},
|
1696 |
-
{
|
1697 |
-
"epoch": 0.39,
|
1698 |
-
"learning_rate": 4.853559627515771e-06,
|
1699 |
-
"loss": 2.76,
|
1700 |
-
"step": 1395
|
1701 |
-
},
|
1702 |
-
{
|
1703 |
-
"epoch": 0.39,
|
1704 |
-
"learning_rate": 4.851682186842896e-06,
|
1705 |
-
"loss": 2.7712,
|
1706 |
-
"step": 1400
|
1707 |
-
},
|
1708 |
-
{
|
1709 |
-
"epoch": 0.39,
|
1710 |
-
"learning_rate": 4.849804746170021e-06,
|
1711 |
-
"loss": 2.8082,
|
1712 |
-
"step": 1405
|
1713 |
-
},
|
1714 |
-
{
|
1715 |
-
"epoch": 0.39,
|
1716 |
-
"learning_rate": 4.8479273054971466e-06,
|
1717 |
-
"loss": 2.9101,
|
1718 |
-
"step": 1410
|
1719 |
-
},
|
1720 |
-
{
|
1721 |
-
"epoch": 0.4,
|
1722 |
-
"learning_rate": 4.846049864824272e-06,
|
1723 |
-
"loss": 2.8121,
|
1724 |
-
"step": 1415
|
1725 |
-
},
|
1726 |
-
{
|
1727 |
-
"epoch": 0.4,
|
1728 |
-
"learning_rate": 4.844172424151397e-06,
|
1729 |
-
"loss": 2.7971,
|
1730 |
-
"step": 1420
|
1731 |
-
},
|
1732 |
-
{
|
1733 |
-
"epoch": 0.4,
|
1734 |
-
"learning_rate": 4.842294983478522e-06,
|
1735 |
-
"loss": 2.8376,
|
1736 |
-
"step": 1425
|
1737 |
-
},
|
1738 |
-
{
|
1739 |
-
"epoch": 0.4,
|
1740 |
-
"learning_rate": 4.8404175428056475e-06,
|
1741 |
-
"loss": 2.7967,
|
1742 |
-
"step": 1430
|
1743 |
-
},
|
1744 |
-
{
|
1745 |
-
"epoch": 0.4,
|
1746 |
-
"learning_rate": 4.838540102132774e-06,
|
1747 |
-
"loss": 2.8066,
|
1748 |
-
"step": 1435
|
1749 |
-
},
|
1750 |
-
{
|
1751 |
-
"epoch": 0.4,
|
1752 |
-
"learning_rate": 4.836662661459898e-06,
|
1753 |
-
"loss": 2.7832,
|
1754 |
-
"step": 1440
|
1755 |
-
},
|
1756 |
-
{
|
1757 |
-
"epoch": 0.4,
|
1758 |
-
"learning_rate": 4.834785220787023e-06,
|
1759 |
-
"loss": 2.7981,
|
1760 |
-
"step": 1445
|
1761 |
-
},
|
1762 |
-
{
|
1763 |
-
"epoch": 0.41,
|
1764 |
-
"learning_rate": 4.8329077801141485e-06,
|
1765 |
-
"loss": 2.743,
|
1766 |
-
"step": 1450
|
1767 |
-
},
|
1768 |
-
{
|
1769 |
-
"epoch": 0.41,
|
1770 |
-
"learning_rate": 4.831030339441274e-06,
|
1771 |
-
"loss": 2.8466,
|
1772 |
-
"step": 1455
|
1773 |
-
},
|
1774 |
-
{
|
1775 |
-
"epoch": 0.41,
|
1776 |
-
"learning_rate": 4.8291528987684e-06,
|
1777 |
-
"loss": 2.8222,
|
1778 |
-
"step": 1460
|
1779 |
-
},
|
1780 |
-
{
|
1781 |
-
"epoch": 0.41,
|
1782 |
-
"learning_rate": 4.827275458095524e-06,
|
1783 |
-
"loss": 2.802,
|
1784 |
-
"step": 1465
|
1785 |
-
},
|
1786 |
-
{
|
1787 |
-
"epoch": 0.41,
|
1788 |
-
"learning_rate": 4.82539801742265e-06,
|
1789 |
-
"loss": 2.6981,
|
1790 |
-
"step": 1470
|
1791 |
-
},
|
1792 |
-
{
|
1793 |
-
"epoch": 0.41,
|
1794 |
-
"learning_rate": 4.823520576749775e-06,
|
1795 |
-
"loss": 2.7656,
|
1796 |
-
"step": 1475
|
1797 |
-
},
|
1798 |
-
{
|
1799 |
-
"epoch": 0.41,
|
1800 |
-
"learning_rate": 4.821643136076901e-06,
|
1801 |
-
"loss": 2.6938,
|
1802 |
-
"step": 1480
|
1803 |
-
},
|
1804 |
-
{
|
1805 |
-
"epoch": 0.41,
|
1806 |
-
"learning_rate": 4.819765695404026e-06,
|
1807 |
-
"loss": 2.7163,
|
1808 |
-
"step": 1485
|
1809 |
-
},
|
1810 |
-
{
|
1811 |
-
"epoch": 0.42,
|
1812 |
-
"learning_rate": 4.81788825473115e-06,
|
1813 |
-
"loss": 2.7057,
|
1814 |
-
"step": 1490
|
1815 |
-
},
|
1816 |
-
{
|
1817 |
-
"epoch": 0.42,
|
1818 |
-
"learning_rate": 4.8160108140582765e-06,
|
1819 |
-
"loss": 2.7461,
|
1820 |
-
"step": 1495
|
1821 |
-
},
|
1822 |
-
{
|
1823 |
-
"epoch": 0.42,
|
1824 |
-
"learning_rate": 4.814133373385401e-06,
|
1825 |
-
"loss": 2.7789,
|
1826 |
-
"step": 1500
|
1827 |
-
},
|
1828 |
-
{
|
1829 |
-
"epoch": 0.42,
|
1830 |
-
"eval_loss": 2.633471965789795,
|
1831 |
-
"eval_runtime": 154.8101,
|
1832 |
-
"eval_samples_per_second": 7.713,
|
1833 |
-
"eval_steps_per_second": 0.969,
|
1834 |
-
"eval_wer": 1.0,
|
1835 |
-
"step": 1500
|
1836 |
-
}
|
1837 |
-
],
|
1838 |
-
"max_steps": 14316,
|
1839 |
-
"num_train_epochs": 4,
|
1840 |
-
"total_flos": 2.1018630215006784e+17,
|
1841 |
-
"trial_name": null,
|
1842 |
-
"trial_params": null
|
1843 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-1500/training_args.bin
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:6f05c6a97d90fd87e90a83547a454ff788a066d3d434f2903b0a9058eae5eb5d
|
3 |
-
size 2991
|
|
|
|
|
|
|
|
checkpoint-2000/config.json
DELETED
@@ -1,109 +0,0 @@
|
|
1 |
-
{
|
2 |
-
"_name_or_path": "facebook/wav2vec2-base-it-voxpopuli",
|
3 |
-
"activation_dropout": 0.1,
|
4 |
-
"adapter_kernel_size": 3,
|
5 |
-
"adapter_stride": 2,
|
6 |
-
"add_adapter": false,
|
7 |
-
"apply_spec_augment": true,
|
8 |
-
"architectures": [
|
9 |
-
"Wav2Vec2ForCTC"
|
10 |
-
],
|
11 |
-
"attention_dropout": 0.1,
|
12 |
-
"bos_token_id": 1,
|
13 |
-
"classifier_proj_size": 256,
|
14 |
-
"codevector_dim": 256,
|
15 |
-
"contrastive_logits_temperature": 0.1,
|
16 |
-
"conv_bias": false,
|
17 |
-
"conv_dim": [
|
18 |
-
512,
|
19 |
-
512,
|
20 |
-
512,
|
21 |
-
512,
|
22 |
-
512,
|
23 |
-
512,
|
24 |
-
512
|
25 |
-
],
|
26 |
-
"conv_kernel": [
|
27 |
-
10,
|
28 |
-
3,
|
29 |
-
3,
|
30 |
-
3,
|
31 |
-
3,
|
32 |
-
2,
|
33 |
-
2
|
34 |
-
],
|
35 |
-
"conv_stride": [
|
36 |
-
5,
|
37 |
-
2,
|
38 |
-
2,
|
39 |
-
2,
|
40 |
-
2,
|
41 |
-
2,
|
42 |
-
2
|
43 |
-
],
|
44 |
-
"ctc_loss_reduction": "mean",
|
45 |
-
"ctc_zero_infinity": false,
|
46 |
-
"diversity_loss_weight": 0.1,
|
47 |
-
"do_stable_layer_norm": false,
|
48 |
-
"eos_token_id": 2,
|
49 |
-
"feat_extract_activation": "gelu",
|
50 |
-
"feat_extract_dropout": 0.0,
|
51 |
-
"feat_extract_norm": "group",
|
52 |
-
"feat_proj_dropout": 0.1,
|
53 |
-
"feat_quantizer_dropout": 0.0,
|
54 |
-
"final_dropout": 0.1,
|
55 |
-
"gradient_checkpointing": false,
|
56 |
-
"hidden_act": "gelu",
|
57 |
-
"hidden_dropout": 0.1,
|
58 |
-
"hidden_dropout_prob": 0.1,
|
59 |
-
"hidden_size": 768,
|
60 |
-
"initializer_range": 0.02,
|
61 |
-
"intermediate_size": 3072,
|
62 |
-
"layer_norm_eps": 1e-05,
|
63 |
-
"layerdrop": 0.1,
|
64 |
-
"mask_feature_length": 10,
|
65 |
-
"mask_feature_min_masks": 0,
|
66 |
-
"mask_feature_prob": 0.0,
|
67 |
-
"mask_time_length": 10,
|
68 |
-
"mask_time_min_masks": 2,
|
69 |
-
"mask_time_prob": 0.05,
|
70 |
-
"model_type": "wav2vec2",
|
71 |
-
"num_adapter_layers": 3,
|
72 |
-
"num_attention_heads": 12,
|
73 |
-
"num_codevector_groups": 2,
|
74 |
-
"num_codevectors_per_group": 320,
|
75 |
-
"num_conv_pos_embedding_groups": 16,
|
76 |
-
"num_conv_pos_embeddings": 128,
|
77 |
-
"num_feat_extract_layers": 7,
|
78 |
-
"num_hidden_layers": 12,
|
79 |
-
"num_negatives": 100,
|
80 |
-
"output_hidden_size": 768,
|
81 |
-
"pad_token_id": 25,
|
82 |
-
"proj_codevector_dim": 256,
|
83 |
-
"tdnn_dilation": [
|
84 |
-
1,
|
85 |
-
2,
|
86 |
-
3,
|
87 |
-
1,
|
88 |
-
1
|
89 |
-
],
|
90 |
-
"tdnn_dim": [
|
91 |
-
512,
|
92 |
-
512,
|
93 |
-
512,
|
94 |
-
512,
|
95 |
-
1500
|
96 |
-
],
|
97 |
-
"tdnn_kernel": [
|
98 |
-
5,
|
99 |
-
3,
|
100 |
-
3,
|
101 |
-
1,
|
102 |
-
1
|
103 |
-
],
|
104 |
-
"torch_dtype": "float32",
|
105 |
-
"transformers_version": "4.16.1",
|
106 |
-
"use_weighted_layer_sum": false,
|
107 |
-
"vocab_size": 32,
|
108 |
-
"xvector_output_dim": 512
|
109 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-2000/optimizer.pt
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:32a02fe3a48cb8f108292c24e359935169139497ac2ce1a31dd7ebe631cc860a
|
3 |
-
size 721685265
|
|
|
|
|
|
|
|
checkpoint-2000/preprocessor_config.json
DELETED
@@ -1,9 +0,0 @@
|
|
1 |
-
{
|
2 |
-
"do_normalize": true,
|
3 |
-
"feature_extractor_type": "Wav2Vec2FeatureExtractor",
|
4 |
-
"feature_size": 1,
|
5 |
-
"padding_side": "right",
|
6 |
-
"padding_value": 0.0,
|
7 |
-
"return_attention_mask": false,
|
8 |
-
"sampling_rate": 16000
|
9 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-2000/pytorch_model.bin
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:ceee4d99fac0e428a3e6c42cf3d28b660b22a63293d37a437f057bc00425c96e
|
3 |
-
size 377670039
|
|
|
|
|
|
|
|
checkpoint-2000/rng_state.pth
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:8cf469464ced309b82e7569855461608a70adf8c27e2aca970016a751298af6e
|
3 |
-
size 14503
|
|
|
|
|
|
|
|
checkpoint-2000/scaler.pt
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:209c0328055994bede183a9757b67dd653e3674aa457a155edc49f9c54d5c4c4
|
3 |
-
size 559
|
|
|
|
|
|
|
|
checkpoint-2000/scheduler.pt
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:98620f310b72cdbfd5eae5ee0c8d95e77ec17d00770e12c509c6b9232279a6de
|
3 |
-
size 623
|
|
|
|
|
|
|
|
checkpoint-2000/trainer_state.json
DELETED
@@ -1,2452 +0,0 @@
|
|
1 |
-
{
|
2 |
-
"best_metric": null,
|
3 |
-
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0.5588153115395362,
|
5 |
-
"global_step": 2000,
|
6 |
-
"is_hyper_param_search": false,
|
7 |
-
"is_local_process_zero": true,
|
8 |
-
"is_world_process_zero": true,
|
9 |
-
"log_history": [
|
10 |
-
{
|
11 |
-
"epoch": 0.0,
|
12 |
-
"learning_rate": 1e-08,
|
13 |
-
"loss": 24.3697,
|
14 |
-
"step": 5
|
15 |
-
},
|
16 |
-
{
|
17 |
-
"epoch": 0.0,
|
18 |
-
"learning_rate": 3.0000000000000004e-08,
|
19 |
-
"loss": 17.6279,
|
20 |
-
"step": 10
|
21 |
-
},
|
22 |
-
{
|
23 |
-
"epoch": 0.0,
|
24 |
-
"learning_rate": 5.5e-08,
|
25 |
-
"loss": 17.3553,
|
26 |
-
"step": 15
|
27 |
-
},
|
28 |
-
{
|
29 |
-
"epoch": 0.01,
|
30 |
-
"learning_rate": 8e-08,
|
31 |
-
"loss": 16.801,
|
32 |
-
"step": 20
|
33 |
-
},
|
34 |
-
{
|
35 |
-
"epoch": 0.01,
|
36 |
-
"learning_rate": 1.0500000000000001e-07,
|
37 |
-
"loss": 15.1858,
|
38 |
-
"step": 25
|
39 |
-
},
|
40 |
-
{
|
41 |
-
"epoch": 0.01,
|
42 |
-
"learning_rate": 1.3e-07,
|
43 |
-
"loss": 14.8524,
|
44 |
-
"step": 30
|
45 |
-
},
|
46 |
-
{
|
47 |
-
"epoch": 0.01,
|
48 |
-
"learning_rate": 1.5500000000000002e-07,
|
49 |
-
"loss": 14.8696,
|
50 |
-
"step": 35
|
51 |
-
},
|
52 |
-
{
|
53 |
-
"epoch": 0.01,
|
54 |
-
"learning_rate": 1.8e-07,
|
55 |
-
"loss": 14.5785,
|
56 |
-
"step": 40
|
57 |
-
},
|
58 |
-
{
|
59 |
-
"epoch": 0.01,
|
60 |
-
"learning_rate": 2.0500000000000002e-07,
|
61 |
-
"loss": 15.0479,
|
62 |
-
"step": 45
|
63 |
-
},
|
64 |
-
{
|
65 |
-
"epoch": 0.01,
|
66 |
-
"learning_rate": 2.3000000000000002e-07,
|
67 |
-
"loss": 14.5691,
|
68 |
-
"step": 50
|
69 |
-
},
|
70 |
-
{
|
71 |
-
"epoch": 0.02,
|
72 |
-
"learning_rate": 2.55e-07,
|
73 |
-
"loss": 18.4354,
|
74 |
-
"step": 55
|
75 |
-
},
|
76 |
-
{
|
77 |
-
"epoch": 0.02,
|
78 |
-
"learning_rate": 2.8e-07,
|
79 |
-
"loss": 18.1091,
|
80 |
-
"step": 60
|
81 |
-
},
|
82 |
-
{
|
83 |
-
"epoch": 0.02,
|
84 |
-
"learning_rate": 3.0500000000000004e-07,
|
85 |
-
"loss": 17.4985,
|
86 |
-
"step": 65
|
87 |
-
},
|
88 |
-
{
|
89 |
-
"epoch": 0.02,
|
90 |
-
"learning_rate": 3.3e-07,
|
91 |
-
"loss": 15.3995,
|
92 |
-
"step": 70
|
93 |
-
},
|
94 |
-
{
|
95 |
-
"epoch": 0.02,
|
96 |
-
"learning_rate": 3.55e-07,
|
97 |
-
"loss": 15.2441,
|
98 |
-
"step": 75
|
99 |
-
},
|
100 |
-
{
|
101 |
-
"epoch": 0.02,
|
102 |
-
"learning_rate": 3.8e-07,
|
103 |
-
"loss": 14.5564,
|
104 |
-
"step": 80
|
105 |
-
},
|
106 |
-
{
|
107 |
-
"epoch": 0.02,
|
108 |
-
"learning_rate": 4.0500000000000004e-07,
|
109 |
-
"loss": 14.824,
|
110 |
-
"step": 85
|
111 |
-
},
|
112 |
-
{
|
113 |
-
"epoch": 0.03,
|
114 |
-
"learning_rate": 4.3e-07,
|
115 |
-
"loss": 14.9045,
|
116 |
-
"step": 90
|
117 |
-
},
|
118 |
-
{
|
119 |
-
"epoch": 0.03,
|
120 |
-
"learning_rate": 4.5500000000000004e-07,
|
121 |
-
"loss": 14.4084,
|
122 |
-
"step": 95
|
123 |
-
},
|
124 |
-
{
|
125 |
-
"epoch": 0.03,
|
126 |
-
"learning_rate": 4.800000000000001e-07,
|
127 |
-
"loss": 14.5942,
|
128 |
-
"step": 100
|
129 |
-
},
|
130 |
-
{
|
131 |
-
"epoch": 0.03,
|
132 |
-
"learning_rate": 5.05e-07,
|
133 |
-
"loss": 18.0525,
|
134 |
-
"step": 105
|
135 |
-
},
|
136 |
-
{
|
137 |
-
"epoch": 0.03,
|
138 |
-
"learning_rate": 5.3e-07,
|
139 |
-
"loss": 16.5825,
|
140 |
-
"step": 110
|
141 |
-
},
|
142 |
-
{
|
143 |
-
"epoch": 0.03,
|
144 |
-
"learning_rate": 5.550000000000001e-07,
|
145 |
-
"loss": 17.0772,
|
146 |
-
"step": 115
|
147 |
-
},
|
148 |
-
{
|
149 |
-
"epoch": 0.03,
|
150 |
-
"learning_rate": 5.800000000000001e-07,
|
151 |
-
"loss": 15.6476,
|
152 |
-
"step": 120
|
153 |
-
},
|
154 |
-
{
|
155 |
-
"epoch": 0.03,
|
156 |
-
"learning_rate": 6.05e-07,
|
157 |
-
"loss": 14.2361,
|
158 |
-
"step": 125
|
159 |
-
},
|
160 |
-
{
|
161 |
-
"epoch": 0.04,
|
162 |
-
"learning_rate": 6.3e-07,
|
163 |
-
"loss": 14.3617,
|
164 |
-
"step": 130
|
165 |
-
},
|
166 |
-
{
|
167 |
-
"epoch": 0.04,
|
168 |
-
"learning_rate": 6.550000000000001e-07,
|
169 |
-
"loss": 14.0812,
|
170 |
-
"step": 135
|
171 |
-
},
|
172 |
-
{
|
173 |
-
"epoch": 0.04,
|
174 |
-
"learning_rate": 6.800000000000001e-07,
|
175 |
-
"loss": 13.7279,
|
176 |
-
"step": 140
|
177 |
-
},
|
178 |
-
{
|
179 |
-
"epoch": 0.04,
|
180 |
-
"learning_rate": 7.05e-07,
|
181 |
-
"loss": 13.5363,
|
182 |
-
"step": 145
|
183 |
-
},
|
184 |
-
{
|
185 |
-
"epoch": 0.04,
|
186 |
-
"learning_rate": 7.3e-07,
|
187 |
-
"loss": 14.8066,
|
188 |
-
"step": 150
|
189 |
-
},
|
190 |
-
{
|
191 |
-
"epoch": 0.04,
|
192 |
-
"learning_rate": 7.5e-07,
|
193 |
-
"loss": 17.2667,
|
194 |
-
"step": 155
|
195 |
-
},
|
196 |
-
{
|
197 |
-
"epoch": 0.04,
|
198 |
-
"learning_rate": 7.750000000000001e-07,
|
199 |
-
"loss": 15.1567,
|
200 |
-
"step": 160
|
201 |
-
},
|
202 |
-
{
|
203 |
-
"epoch": 0.05,
|
204 |
-
"learning_rate": 8.000000000000001e-07,
|
205 |
-
"loss": 16.3138,
|
206 |
-
"step": 165
|
207 |
-
},
|
208 |
-
{
|
209 |
-
"epoch": 0.05,
|
210 |
-
"learning_rate": 8.250000000000001e-07,
|
211 |
-
"loss": 13.1312,
|
212 |
-
"step": 170
|
213 |
-
},
|
214 |
-
{
|
215 |
-
"epoch": 0.05,
|
216 |
-
"learning_rate": 8.500000000000001e-07,
|
217 |
-
"loss": 12.7545,
|
218 |
-
"step": 175
|
219 |
-
},
|
220 |
-
{
|
221 |
-
"epoch": 0.05,
|
222 |
-
"learning_rate": 8.75e-07,
|
223 |
-
"loss": 13.0234,
|
224 |
-
"step": 180
|
225 |
-
},
|
226 |
-
{
|
227 |
-
"epoch": 0.05,
|
228 |
-
"learning_rate": 9.000000000000001e-07,
|
229 |
-
"loss": 13.2179,
|
230 |
-
"step": 185
|
231 |
-
},
|
232 |
-
{
|
233 |
-
"epoch": 0.05,
|
234 |
-
"learning_rate": 9.25e-07,
|
235 |
-
"loss": 13.0916,
|
236 |
-
"step": 190
|
237 |
-
},
|
238 |
-
{
|
239 |
-
"epoch": 0.05,
|
240 |
-
"learning_rate": 9.500000000000001e-07,
|
241 |
-
"loss": 11.6765,
|
242 |
-
"step": 195
|
243 |
-
},
|
244 |
-
{
|
245 |
-
"epoch": 0.06,
|
246 |
-
"learning_rate": 9.750000000000002e-07,
|
247 |
-
"loss": 12.5044,
|
248 |
-
"step": 200
|
249 |
-
},
|
250 |
-
{
|
251 |
-
"epoch": 0.06,
|
252 |
-
"learning_rate": 1.0000000000000002e-06,
|
253 |
-
"loss": 15.0522,
|
254 |
-
"step": 205
|
255 |
-
},
|
256 |
-
{
|
257 |
-
"epoch": 0.06,
|
258 |
-
"learning_rate": 1.025e-06,
|
259 |
-
"loss": 12.4961,
|
260 |
-
"step": 210
|
261 |
-
},
|
262 |
-
{
|
263 |
-
"epoch": 0.06,
|
264 |
-
"learning_rate": 1.0500000000000001e-06,
|
265 |
-
"loss": 11.6473,
|
266 |
-
"step": 215
|
267 |
-
},
|
268 |
-
{
|
269 |
-
"epoch": 0.06,
|
270 |
-
"learning_rate": 1.075e-06,
|
271 |
-
"loss": 10.5303,
|
272 |
-
"step": 220
|
273 |
-
},
|
274 |
-
{
|
275 |
-
"epoch": 0.06,
|
276 |
-
"learning_rate": 1.1e-06,
|
277 |
-
"loss": 12.0959,
|
278 |
-
"step": 225
|
279 |
-
},
|
280 |
-
{
|
281 |
-
"epoch": 0.06,
|
282 |
-
"learning_rate": 1.125e-06,
|
283 |
-
"loss": 10.7053,
|
284 |
-
"step": 230
|
285 |
-
},
|
286 |
-
{
|
287 |
-
"epoch": 0.07,
|
288 |
-
"learning_rate": 1.1500000000000002e-06,
|
289 |
-
"loss": 9.5165,
|
290 |
-
"step": 235
|
291 |
-
},
|
292 |
-
{
|
293 |
-
"epoch": 0.07,
|
294 |
-
"learning_rate": 1.175e-06,
|
295 |
-
"loss": 11.838,
|
296 |
-
"step": 240
|
297 |
-
},
|
298 |
-
{
|
299 |
-
"epoch": 0.07,
|
300 |
-
"learning_rate": 1.2000000000000002e-06,
|
301 |
-
"loss": 9.5919,
|
302 |
-
"step": 245
|
303 |
-
},
|
304 |
-
{
|
305 |
-
"epoch": 0.07,
|
306 |
-
"learning_rate": 1.2250000000000001e-06,
|
307 |
-
"loss": 9.7433,
|
308 |
-
"step": 250
|
309 |
-
},
|
310 |
-
{
|
311 |
-
"epoch": 0.07,
|
312 |
-
"learning_rate": 1.25e-06,
|
313 |
-
"loss": 10.6129,
|
314 |
-
"step": 255
|
315 |
-
},
|
316 |
-
{
|
317 |
-
"epoch": 0.07,
|
318 |
-
"learning_rate": 1.275e-06,
|
319 |
-
"loss": 9.6936,
|
320 |
-
"step": 260
|
321 |
-
},
|
322 |
-
{
|
323 |
-
"epoch": 0.07,
|
324 |
-
"learning_rate": 1.3e-06,
|
325 |
-
"loss": 10.8679,
|
326 |
-
"step": 265
|
327 |
-
},
|
328 |
-
{
|
329 |
-
"epoch": 0.08,
|
330 |
-
"learning_rate": 1.3250000000000002e-06,
|
331 |
-
"loss": 12.5521,
|
332 |
-
"step": 270
|
333 |
-
},
|
334 |
-
{
|
335 |
-
"epoch": 0.08,
|
336 |
-
"learning_rate": 1.3500000000000002e-06,
|
337 |
-
"loss": 8.2902,
|
338 |
-
"step": 275
|
339 |
-
},
|
340 |
-
{
|
341 |
-
"epoch": 0.08,
|
342 |
-
"learning_rate": 1.3750000000000002e-06,
|
343 |
-
"loss": 8.7901,
|
344 |
-
"step": 280
|
345 |
-
},
|
346 |
-
{
|
347 |
-
"epoch": 0.08,
|
348 |
-
"learning_rate": 1.4000000000000001e-06,
|
349 |
-
"loss": 7.9864,
|
350 |
-
"step": 285
|
351 |
-
},
|
352 |
-
{
|
353 |
-
"epoch": 0.08,
|
354 |
-
"learning_rate": 1.425e-06,
|
355 |
-
"loss": 8.0575,
|
356 |
-
"step": 290
|
357 |
-
},
|
358 |
-
{
|
359 |
-
"epoch": 0.08,
|
360 |
-
"learning_rate": 1.45e-06,
|
361 |
-
"loss": 7.4836,
|
362 |
-
"step": 295
|
363 |
-
},
|
364 |
-
{
|
365 |
-
"epoch": 0.08,
|
366 |
-
"learning_rate": 1.475e-06,
|
367 |
-
"loss": 10.0019,
|
368 |
-
"step": 300
|
369 |
-
},
|
370 |
-
{
|
371 |
-
"epoch": 0.09,
|
372 |
-
"learning_rate": 1.5e-06,
|
373 |
-
"loss": 11.417,
|
374 |
-
"step": 305
|
375 |
-
},
|
376 |
-
{
|
377 |
-
"epoch": 0.09,
|
378 |
-
"learning_rate": 1.525e-06,
|
379 |
-
"loss": 8.4366,
|
380 |
-
"step": 310
|
381 |
-
},
|
382 |
-
{
|
383 |
-
"epoch": 0.09,
|
384 |
-
"learning_rate": 1.5500000000000002e-06,
|
385 |
-
"loss": 8.6453,
|
386 |
-
"step": 315
|
387 |
-
},
|
388 |
-
{
|
389 |
-
"epoch": 0.09,
|
390 |
-
"learning_rate": 1.5750000000000002e-06,
|
391 |
-
"loss": 7.6421,
|
392 |
-
"step": 320
|
393 |
-
},
|
394 |
-
{
|
395 |
-
"epoch": 0.09,
|
396 |
-
"learning_rate": 1.6000000000000001e-06,
|
397 |
-
"loss": 8.7056,
|
398 |
-
"step": 325
|
399 |
-
},
|
400 |
-
{
|
401 |
-
"epoch": 0.09,
|
402 |
-
"learning_rate": 1.6250000000000001e-06,
|
403 |
-
"loss": 7.4756,
|
404 |
-
"step": 330
|
405 |
-
},
|
406 |
-
{
|
407 |
-
"epoch": 0.09,
|
408 |
-
"learning_rate": 1.6500000000000003e-06,
|
409 |
-
"loss": 6.6272,
|
410 |
-
"step": 335
|
411 |
-
},
|
412 |
-
{
|
413 |
-
"epoch": 0.09,
|
414 |
-
"learning_rate": 1.6750000000000003e-06,
|
415 |
-
"loss": 6.9702,
|
416 |
-
"step": 340
|
417 |
-
},
|
418 |
-
{
|
419 |
-
"epoch": 0.1,
|
420 |
-
"learning_rate": 1.7000000000000002e-06,
|
421 |
-
"loss": 6.971,
|
422 |
-
"step": 345
|
423 |
-
},
|
424 |
-
{
|
425 |
-
"epoch": 0.1,
|
426 |
-
"learning_rate": 1.725e-06,
|
427 |
-
"loss": 6.4695,
|
428 |
-
"step": 350
|
429 |
-
},
|
430 |
-
{
|
431 |
-
"epoch": 0.1,
|
432 |
-
"learning_rate": 1.75e-06,
|
433 |
-
"loss": 9.6089,
|
434 |
-
"step": 355
|
435 |
-
},
|
436 |
-
{
|
437 |
-
"epoch": 0.1,
|
438 |
-
"learning_rate": 1.7750000000000002e-06,
|
439 |
-
"loss": 7.6838,
|
440 |
-
"step": 360
|
441 |
-
},
|
442 |
-
{
|
443 |
-
"epoch": 0.1,
|
444 |
-
"learning_rate": 1.8000000000000001e-06,
|
445 |
-
"loss": 6.6076,
|
446 |
-
"step": 365
|
447 |
-
},
|
448 |
-
{
|
449 |
-
"epoch": 0.1,
|
450 |
-
"learning_rate": 1.825e-06,
|
451 |
-
"loss": 8.1119,
|
452 |
-
"step": 370
|
453 |
-
},
|
454 |
-
{
|
455 |
-
"epoch": 0.1,
|
456 |
-
"learning_rate": 1.85e-06,
|
457 |
-
"loss": 8.6286,
|
458 |
-
"step": 375
|
459 |
-
},
|
460 |
-
{
|
461 |
-
"epoch": 0.11,
|
462 |
-
"learning_rate": 1.8750000000000003e-06,
|
463 |
-
"loss": 6.2768,
|
464 |
-
"step": 380
|
465 |
-
},
|
466 |
-
{
|
467 |
-
"epoch": 0.11,
|
468 |
-
"learning_rate": 1.9000000000000002e-06,
|
469 |
-
"loss": 6.9159,
|
470 |
-
"step": 385
|
471 |
-
},
|
472 |
-
{
|
473 |
-
"epoch": 0.11,
|
474 |
-
"learning_rate": 1.925e-06,
|
475 |
-
"loss": 7.0885,
|
476 |
-
"step": 390
|
477 |
-
},
|
478 |
-
{
|
479 |
-
"epoch": 0.11,
|
480 |
-
"learning_rate": 1.9500000000000004e-06,
|
481 |
-
"loss": 8.5395,
|
482 |
-
"step": 395
|
483 |
-
},
|
484 |
-
{
|
485 |
-
"epoch": 0.11,
|
486 |
-
"learning_rate": 1.975e-06,
|
487 |
-
"loss": 8.7365,
|
488 |
-
"step": 400
|
489 |
-
},
|
490 |
-
{
|
491 |
-
"epoch": 0.11,
|
492 |
-
"learning_rate": 2.0000000000000003e-06,
|
493 |
-
"loss": 8.7254,
|
494 |
-
"step": 405
|
495 |
-
},
|
496 |
-
{
|
497 |
-
"epoch": 0.11,
|
498 |
-
"learning_rate": 2.025e-06,
|
499 |
-
"loss": 6.4878,
|
500 |
-
"step": 410
|
501 |
-
},
|
502 |
-
{
|
503 |
-
"epoch": 0.12,
|
504 |
-
"learning_rate": 2.05e-06,
|
505 |
-
"loss": 10.4269,
|
506 |
-
"step": 415
|
507 |
-
},
|
508 |
-
{
|
509 |
-
"epoch": 0.12,
|
510 |
-
"learning_rate": 2.075e-06,
|
511 |
-
"loss": 6.2613,
|
512 |
-
"step": 420
|
513 |
-
},
|
514 |
-
{
|
515 |
-
"epoch": 0.12,
|
516 |
-
"learning_rate": 2.1000000000000002e-06,
|
517 |
-
"loss": 6.6001,
|
518 |
-
"step": 425
|
519 |
-
},
|
520 |
-
{
|
521 |
-
"epoch": 0.12,
|
522 |
-
"learning_rate": 2.125e-06,
|
523 |
-
"loss": 6.0585,
|
524 |
-
"step": 430
|
525 |
-
},
|
526 |
-
{
|
527 |
-
"epoch": 0.12,
|
528 |
-
"learning_rate": 2.15e-06,
|
529 |
-
"loss": 6.2137,
|
530 |
-
"step": 435
|
531 |
-
},
|
532 |
-
{
|
533 |
-
"epoch": 0.12,
|
534 |
-
"learning_rate": 2.1750000000000004e-06,
|
535 |
-
"loss": 8.026,
|
536 |
-
"step": 440
|
537 |
-
},
|
538 |
-
{
|
539 |
-
"epoch": 0.12,
|
540 |
-
"learning_rate": 2.2e-06,
|
541 |
-
"loss": 8.5144,
|
542 |
-
"step": 445
|
543 |
-
},
|
544 |
-
{
|
545 |
-
"epoch": 0.13,
|
546 |
-
"learning_rate": 2.2250000000000003e-06,
|
547 |
-
"loss": 6.092,
|
548 |
-
"step": 450
|
549 |
-
},
|
550 |
-
{
|
551 |
-
"epoch": 0.13,
|
552 |
-
"learning_rate": 2.25e-06,
|
553 |
-
"loss": 8.8018,
|
554 |
-
"step": 455
|
555 |
-
},
|
556 |
-
{
|
557 |
-
"epoch": 0.13,
|
558 |
-
"learning_rate": 2.2750000000000002e-06,
|
559 |
-
"loss": 6.2074,
|
560 |
-
"step": 460
|
561 |
-
},
|
562 |
-
{
|
563 |
-
"epoch": 0.13,
|
564 |
-
"learning_rate": 2.3000000000000004e-06,
|
565 |
-
"loss": 10.6086,
|
566 |
-
"step": 465
|
567 |
-
},
|
568 |
-
{
|
569 |
-
"epoch": 0.13,
|
570 |
-
"learning_rate": 2.325e-06,
|
571 |
-
"loss": 6.5399,
|
572 |
-
"step": 470
|
573 |
-
},
|
574 |
-
{
|
575 |
-
"epoch": 0.13,
|
576 |
-
"learning_rate": 2.35e-06,
|
577 |
-
"loss": 7.0942,
|
578 |
-
"step": 475
|
579 |
-
},
|
580 |
-
{
|
581 |
-
"epoch": 0.13,
|
582 |
-
"learning_rate": 2.375e-06,
|
583 |
-
"loss": 6.4347,
|
584 |
-
"step": 480
|
585 |
-
},
|
586 |
-
{
|
587 |
-
"epoch": 0.14,
|
588 |
-
"learning_rate": 2.4000000000000003e-06,
|
589 |
-
"loss": 7.1602,
|
590 |
-
"step": 485
|
591 |
-
},
|
592 |
-
{
|
593 |
-
"epoch": 0.14,
|
594 |
-
"learning_rate": 2.425e-06,
|
595 |
-
"loss": 5.4093,
|
596 |
-
"step": 490
|
597 |
-
},
|
598 |
-
{
|
599 |
-
"epoch": 0.14,
|
600 |
-
"learning_rate": 2.4500000000000003e-06,
|
601 |
-
"loss": 6.618,
|
602 |
-
"step": 495
|
603 |
-
},
|
604 |
-
{
|
605 |
-
"epoch": 0.14,
|
606 |
-
"learning_rate": 2.475e-06,
|
607 |
-
"loss": 8.3884,
|
608 |
-
"step": 500
|
609 |
-
},
|
610 |
-
{
|
611 |
-
"epoch": 0.14,
|
612 |
-
"eval_loss": 7.4285359382629395,
|
613 |
-
"eval_runtime": 109.7145,
|
614 |
-
"eval_samples_per_second": 10.883,
|
615 |
-
"eval_steps_per_second": 1.367,
|
616 |
-
"eval_wer": 1.0,
|
617 |
-
"step": 500
|
618 |
-
},
|
619 |
-
{
|
620 |
-
"epoch": 0.14,
|
621 |
-
"learning_rate": 2.5e-06,
|
622 |
-
"loss": 8.5122,
|
623 |
-
"step": 505
|
624 |
-
},
|
625 |
-
{
|
626 |
-
"epoch": 0.14,
|
627 |
-
"learning_rate": 2.5250000000000004e-06,
|
628 |
-
"loss": 9.0237,
|
629 |
-
"step": 510
|
630 |
-
},
|
631 |
-
{
|
632 |
-
"epoch": 0.14,
|
633 |
-
"learning_rate": 2.55e-06,
|
634 |
-
"loss": 6.343,
|
635 |
-
"step": 515
|
636 |
-
},
|
637 |
-
{
|
638 |
-
"epoch": 0.15,
|
639 |
-
"learning_rate": 2.5750000000000003e-06,
|
640 |
-
"loss": 6.3103,
|
641 |
-
"step": 520
|
642 |
-
},
|
643 |
-
{
|
644 |
-
"epoch": 0.15,
|
645 |
-
"learning_rate": 2.6e-06,
|
646 |
-
"loss": 7.6124,
|
647 |
-
"step": 525
|
648 |
-
},
|
649 |
-
{
|
650 |
-
"epoch": 0.15,
|
651 |
-
"learning_rate": 2.6250000000000003e-06,
|
652 |
-
"loss": 7.1252,
|
653 |
-
"step": 530
|
654 |
-
},
|
655 |
-
{
|
656 |
-
"epoch": 0.15,
|
657 |
-
"learning_rate": 2.6500000000000005e-06,
|
658 |
-
"loss": 7.0198,
|
659 |
-
"step": 535
|
660 |
-
},
|
661 |
-
{
|
662 |
-
"epoch": 0.15,
|
663 |
-
"learning_rate": 2.6750000000000002e-06,
|
664 |
-
"loss": 6.3734,
|
665 |
-
"step": 540
|
666 |
-
},
|
667 |
-
{
|
668 |
-
"epoch": 0.15,
|
669 |
-
"learning_rate": 2.7000000000000004e-06,
|
670 |
-
"loss": 4.7063,
|
671 |
-
"step": 545
|
672 |
-
},
|
673 |
-
{
|
674 |
-
"epoch": 0.15,
|
675 |
-
"learning_rate": 2.7250000000000006e-06,
|
676 |
-
"loss": 7.7081,
|
677 |
-
"step": 550
|
678 |
-
},
|
679 |
-
{
|
680 |
-
"epoch": 0.16,
|
681 |
-
"learning_rate": 2.7500000000000004e-06,
|
682 |
-
"loss": 5.9,
|
683 |
-
"step": 555
|
684 |
-
},
|
685 |
-
{
|
686 |
-
"epoch": 0.16,
|
687 |
-
"learning_rate": 2.7750000000000005e-06,
|
688 |
-
"loss": 5.634,
|
689 |
-
"step": 560
|
690 |
-
},
|
691 |
-
{
|
692 |
-
"epoch": 0.16,
|
693 |
-
"learning_rate": 2.8000000000000003e-06,
|
694 |
-
"loss": 8.6987,
|
695 |
-
"step": 565
|
696 |
-
},
|
697 |
-
{
|
698 |
-
"epoch": 0.16,
|
699 |
-
"learning_rate": 2.825e-06,
|
700 |
-
"loss": 6.5511,
|
701 |
-
"step": 570
|
702 |
-
},
|
703 |
-
{
|
704 |
-
"epoch": 0.16,
|
705 |
-
"learning_rate": 2.85e-06,
|
706 |
-
"loss": 7.6908,
|
707 |
-
"step": 575
|
708 |
-
},
|
709 |
-
{
|
710 |
-
"epoch": 0.16,
|
711 |
-
"learning_rate": 2.875e-06,
|
712 |
-
"loss": 7.4105,
|
713 |
-
"step": 580
|
714 |
-
},
|
715 |
-
{
|
716 |
-
"epoch": 0.16,
|
717 |
-
"learning_rate": 2.9e-06,
|
718 |
-
"loss": 5.802,
|
719 |
-
"step": 585
|
720 |
-
},
|
721 |
-
{
|
722 |
-
"epoch": 0.16,
|
723 |
-
"learning_rate": 2.925e-06,
|
724 |
-
"loss": 6.0835,
|
725 |
-
"step": 590
|
726 |
-
},
|
727 |
-
{
|
728 |
-
"epoch": 0.17,
|
729 |
-
"learning_rate": 2.95e-06,
|
730 |
-
"loss": 7.8744,
|
731 |
-
"step": 595
|
732 |
-
},
|
733 |
-
{
|
734 |
-
"epoch": 0.17,
|
735 |
-
"learning_rate": 2.9750000000000003e-06,
|
736 |
-
"loss": 6.1134,
|
737 |
-
"step": 600
|
738 |
-
},
|
739 |
-
{
|
740 |
-
"epoch": 0.17,
|
741 |
-
"learning_rate": 3e-06,
|
742 |
-
"loss": 6.6416,
|
743 |
-
"step": 605
|
744 |
-
},
|
745 |
-
{
|
746 |
-
"epoch": 0.17,
|
747 |
-
"learning_rate": 3.0250000000000003e-06,
|
748 |
-
"loss": 8.7177,
|
749 |
-
"step": 610
|
750 |
-
},
|
751 |
-
{
|
752 |
-
"epoch": 0.17,
|
753 |
-
"learning_rate": 3.05e-06,
|
754 |
-
"loss": 6.2383,
|
755 |
-
"step": 615
|
756 |
-
},
|
757 |
-
{
|
758 |
-
"epoch": 0.17,
|
759 |
-
"learning_rate": 3.075e-06,
|
760 |
-
"loss": 6.2413,
|
761 |
-
"step": 620
|
762 |
-
},
|
763 |
-
{
|
764 |
-
"epoch": 0.17,
|
765 |
-
"learning_rate": 3.1000000000000004e-06,
|
766 |
-
"loss": 5.4088,
|
767 |
-
"step": 625
|
768 |
-
},
|
769 |
-
{
|
770 |
-
"epoch": 0.18,
|
771 |
-
"learning_rate": 3.125e-06,
|
772 |
-
"loss": 4.3896,
|
773 |
-
"step": 630
|
774 |
-
},
|
775 |
-
{
|
776 |
-
"epoch": 0.18,
|
777 |
-
"learning_rate": 3.1500000000000003e-06,
|
778 |
-
"loss": 5.6138,
|
779 |
-
"step": 635
|
780 |
-
},
|
781 |
-
{
|
782 |
-
"epoch": 0.18,
|
783 |
-
"learning_rate": 3.175e-06,
|
784 |
-
"loss": 5.0078,
|
785 |
-
"step": 640
|
786 |
-
},
|
787 |
-
{
|
788 |
-
"epoch": 0.18,
|
789 |
-
"learning_rate": 3.2000000000000003e-06,
|
790 |
-
"loss": 6.9994,
|
791 |
-
"step": 645
|
792 |
-
},
|
793 |
-
{
|
794 |
-
"epoch": 0.18,
|
795 |
-
"learning_rate": 3.2250000000000005e-06,
|
796 |
-
"loss": 5.4662,
|
797 |
-
"step": 650
|
798 |
-
},
|
799 |
-
{
|
800 |
-
"epoch": 0.18,
|
801 |
-
"learning_rate": 3.2500000000000002e-06,
|
802 |
-
"loss": 5.144,
|
803 |
-
"step": 655
|
804 |
-
},
|
805 |
-
{
|
806 |
-
"epoch": 0.18,
|
807 |
-
"learning_rate": 3.2750000000000004e-06,
|
808 |
-
"loss": 5.4199,
|
809 |
-
"step": 660
|
810 |
-
},
|
811 |
-
{
|
812 |
-
"epoch": 0.19,
|
813 |
-
"learning_rate": 3.3000000000000006e-06,
|
814 |
-
"loss": 6.5677,
|
815 |
-
"step": 665
|
816 |
-
},
|
817 |
-
{
|
818 |
-
"epoch": 0.19,
|
819 |
-
"learning_rate": 3.3250000000000004e-06,
|
820 |
-
"loss": 6.168,
|
821 |
-
"step": 670
|
822 |
-
},
|
823 |
-
{
|
824 |
-
"epoch": 0.19,
|
825 |
-
"learning_rate": 3.3500000000000005e-06,
|
826 |
-
"loss": 6.3248,
|
827 |
-
"step": 675
|
828 |
-
},
|
829 |
-
{
|
830 |
-
"epoch": 0.19,
|
831 |
-
"learning_rate": 3.3750000000000003e-06,
|
832 |
-
"loss": 5.6306,
|
833 |
-
"step": 680
|
834 |
-
},
|
835 |
-
{
|
836 |
-
"epoch": 0.19,
|
837 |
-
"learning_rate": 3.4000000000000005e-06,
|
838 |
-
"loss": 4.4895,
|
839 |
-
"step": 685
|
840 |
-
},
|
841 |
-
{
|
842 |
-
"epoch": 0.19,
|
843 |
-
"learning_rate": 3.4250000000000007e-06,
|
844 |
-
"loss": 4.7549,
|
845 |
-
"step": 690
|
846 |
-
},
|
847 |
-
{
|
848 |
-
"epoch": 0.19,
|
849 |
-
"learning_rate": 3.45e-06,
|
850 |
-
"loss": 3.8521,
|
851 |
-
"step": 695
|
852 |
-
},
|
853 |
-
{
|
854 |
-
"epoch": 0.2,
|
855 |
-
"learning_rate": 3.475e-06,
|
856 |
-
"loss": 7.8289,
|
857 |
-
"step": 700
|
858 |
-
},
|
859 |
-
{
|
860 |
-
"epoch": 0.2,
|
861 |
-
"learning_rate": 3.5e-06,
|
862 |
-
"loss": 8.7039,
|
863 |
-
"step": 705
|
864 |
-
},
|
865 |
-
{
|
866 |
-
"epoch": 0.2,
|
867 |
-
"learning_rate": 3.525e-06,
|
868 |
-
"loss": 7.9723,
|
869 |
-
"step": 710
|
870 |
-
},
|
871 |
-
{
|
872 |
-
"epoch": 0.2,
|
873 |
-
"learning_rate": 3.5500000000000003e-06,
|
874 |
-
"loss": 7.4315,
|
875 |
-
"step": 715
|
876 |
-
},
|
877 |
-
{
|
878 |
-
"epoch": 0.2,
|
879 |
-
"learning_rate": 3.575e-06,
|
880 |
-
"loss": 5.6061,
|
881 |
-
"step": 720
|
882 |
-
},
|
883 |
-
{
|
884 |
-
"epoch": 0.2,
|
885 |
-
"learning_rate": 3.6000000000000003e-06,
|
886 |
-
"loss": 5.4848,
|
887 |
-
"step": 725
|
888 |
-
},
|
889 |
-
{
|
890 |
-
"epoch": 0.2,
|
891 |
-
"learning_rate": 3.625e-06,
|
892 |
-
"loss": 4.1425,
|
893 |
-
"step": 730
|
894 |
-
},
|
895 |
-
{
|
896 |
-
"epoch": 0.21,
|
897 |
-
"learning_rate": 3.65e-06,
|
898 |
-
"loss": 5.5031,
|
899 |
-
"step": 735
|
900 |
-
},
|
901 |
-
{
|
902 |
-
"epoch": 0.21,
|
903 |
-
"learning_rate": 3.6750000000000004e-06,
|
904 |
-
"loss": 4.4556,
|
905 |
-
"step": 740
|
906 |
-
},
|
907 |
-
{
|
908 |
-
"epoch": 0.21,
|
909 |
-
"learning_rate": 3.7e-06,
|
910 |
-
"loss": 5.9729,
|
911 |
-
"step": 745
|
912 |
-
},
|
913 |
-
{
|
914 |
-
"epoch": 0.21,
|
915 |
-
"learning_rate": 3.7250000000000003e-06,
|
916 |
-
"loss": 4.5424,
|
917 |
-
"step": 750
|
918 |
-
},
|
919 |
-
{
|
920 |
-
"epoch": 0.21,
|
921 |
-
"learning_rate": 3.7500000000000005e-06,
|
922 |
-
"loss": 7.1968,
|
923 |
-
"step": 755
|
924 |
-
},
|
925 |
-
{
|
926 |
-
"epoch": 0.21,
|
927 |
-
"learning_rate": 3.7750000000000003e-06,
|
928 |
-
"loss": 6.0854,
|
929 |
-
"step": 760
|
930 |
-
},
|
931 |
-
{
|
932 |
-
"epoch": 0.21,
|
933 |
-
"learning_rate": 3.8000000000000005e-06,
|
934 |
-
"loss": 5.1684,
|
935 |
-
"step": 765
|
936 |
-
},
|
937 |
-
{
|
938 |
-
"epoch": 0.22,
|
939 |
-
"learning_rate": 3.825000000000001e-06,
|
940 |
-
"loss": 4.8008,
|
941 |
-
"step": 770
|
942 |
-
},
|
943 |
-
{
|
944 |
-
"epoch": 0.22,
|
945 |
-
"learning_rate": 3.85e-06,
|
946 |
-
"loss": 5.2927,
|
947 |
-
"step": 775
|
948 |
-
},
|
949 |
-
{
|
950 |
-
"epoch": 0.22,
|
951 |
-
"learning_rate": 3.875e-06,
|
952 |
-
"loss": 4.1035,
|
953 |
-
"step": 780
|
954 |
-
},
|
955 |
-
{
|
956 |
-
"epoch": 0.22,
|
957 |
-
"learning_rate": 3.900000000000001e-06,
|
958 |
-
"loss": 4.083,
|
959 |
-
"step": 785
|
960 |
-
},
|
961 |
-
{
|
962 |
-
"epoch": 0.22,
|
963 |
-
"learning_rate": 3.9250000000000005e-06,
|
964 |
-
"loss": 4.2797,
|
965 |
-
"step": 790
|
966 |
-
},
|
967 |
-
{
|
968 |
-
"epoch": 0.22,
|
969 |
-
"learning_rate": 3.95e-06,
|
970 |
-
"loss": 6.4263,
|
971 |
-
"step": 795
|
972 |
-
},
|
973 |
-
{
|
974 |
-
"epoch": 0.22,
|
975 |
-
"learning_rate": 3.975000000000001e-06,
|
976 |
-
"loss": 6.5447,
|
977 |
-
"step": 800
|
978 |
-
},
|
979 |
-
{
|
980 |
-
"epoch": 0.22,
|
981 |
-
"learning_rate": 4.000000000000001e-06,
|
982 |
-
"loss": 5.9181,
|
983 |
-
"step": 805
|
984 |
-
},
|
985 |
-
{
|
986 |
-
"epoch": 0.23,
|
987 |
-
"learning_rate": 4.0250000000000004e-06,
|
988 |
-
"loss": 6.3199,
|
989 |
-
"step": 810
|
990 |
-
},
|
991 |
-
{
|
992 |
-
"epoch": 0.23,
|
993 |
-
"learning_rate": 4.05e-06,
|
994 |
-
"loss": 7.4045,
|
995 |
-
"step": 815
|
996 |
-
},
|
997 |
-
{
|
998 |
-
"epoch": 0.23,
|
999 |
-
"learning_rate": 4.075e-06,
|
1000 |
-
"loss": 4.7224,
|
1001 |
-
"step": 820
|
1002 |
-
},
|
1003 |
-
{
|
1004 |
-
"epoch": 0.23,
|
1005 |
-
"learning_rate": 4.1e-06,
|
1006 |
-
"loss": 4.3196,
|
1007 |
-
"step": 825
|
1008 |
-
},
|
1009 |
-
{
|
1010 |
-
"epoch": 0.23,
|
1011 |
-
"learning_rate": 4.125e-06,
|
1012 |
-
"loss": 6.7817,
|
1013 |
-
"step": 830
|
1014 |
-
},
|
1015 |
-
{
|
1016 |
-
"epoch": 0.23,
|
1017 |
-
"learning_rate": 4.15e-06,
|
1018 |
-
"loss": 3.9509,
|
1019 |
-
"step": 835
|
1020 |
-
},
|
1021 |
-
{
|
1022 |
-
"epoch": 0.23,
|
1023 |
-
"learning_rate": 4.175e-06,
|
1024 |
-
"loss": 5.2207,
|
1025 |
-
"step": 840
|
1026 |
-
},
|
1027 |
-
{
|
1028 |
-
"epoch": 0.24,
|
1029 |
-
"learning_rate": 4.2000000000000004e-06,
|
1030 |
-
"loss": 4.5517,
|
1031 |
-
"step": 845
|
1032 |
-
},
|
1033 |
-
{
|
1034 |
-
"epoch": 0.24,
|
1035 |
-
"learning_rate": 4.225e-06,
|
1036 |
-
"loss": 5.7441,
|
1037 |
-
"step": 850
|
1038 |
-
},
|
1039 |
-
{
|
1040 |
-
"epoch": 0.24,
|
1041 |
-
"learning_rate": 4.25e-06,
|
1042 |
-
"loss": 6.9507,
|
1043 |
-
"step": 855
|
1044 |
-
},
|
1045 |
-
{
|
1046 |
-
"epoch": 0.24,
|
1047 |
-
"learning_rate": 4.2750000000000006e-06,
|
1048 |
-
"loss": 4.8579,
|
1049 |
-
"step": 860
|
1050 |
-
},
|
1051 |
-
{
|
1052 |
-
"epoch": 0.24,
|
1053 |
-
"learning_rate": 4.3e-06,
|
1054 |
-
"loss": 5.6729,
|
1055 |
-
"step": 865
|
1056 |
-
},
|
1057 |
-
{
|
1058 |
-
"epoch": 0.24,
|
1059 |
-
"learning_rate": 4.325e-06,
|
1060 |
-
"loss": 5.7697,
|
1061 |
-
"step": 870
|
1062 |
-
},
|
1063 |
-
{
|
1064 |
-
"epoch": 0.24,
|
1065 |
-
"learning_rate": 4.350000000000001e-06,
|
1066 |
-
"loss": 4.7595,
|
1067 |
-
"step": 875
|
1068 |
-
},
|
1069 |
-
{
|
1070 |
-
"epoch": 0.25,
|
1071 |
-
"learning_rate": 4.3750000000000005e-06,
|
1072 |
-
"loss": 4.6507,
|
1073 |
-
"step": 880
|
1074 |
-
},
|
1075 |
-
{
|
1076 |
-
"epoch": 0.25,
|
1077 |
-
"learning_rate": 4.4e-06,
|
1078 |
-
"loss": 5.0327,
|
1079 |
-
"step": 885
|
1080 |
-
},
|
1081 |
-
{
|
1082 |
-
"epoch": 0.25,
|
1083 |
-
"learning_rate": 4.425e-06,
|
1084 |
-
"loss": 4.3302,
|
1085 |
-
"step": 890
|
1086 |
-
},
|
1087 |
-
{
|
1088 |
-
"epoch": 0.25,
|
1089 |
-
"learning_rate": 4.450000000000001e-06,
|
1090 |
-
"loss": 3.2291,
|
1091 |
-
"step": 895
|
1092 |
-
},
|
1093 |
-
{
|
1094 |
-
"epoch": 0.25,
|
1095 |
-
"learning_rate": 4.475e-06,
|
1096 |
-
"loss": 3.5748,
|
1097 |
-
"step": 900
|
1098 |
-
},
|
1099 |
-
{
|
1100 |
-
"epoch": 0.25,
|
1101 |
-
"learning_rate": 4.5e-06,
|
1102 |
-
"loss": 2.9844,
|
1103 |
-
"step": 905
|
1104 |
-
},
|
1105 |
-
{
|
1106 |
-
"epoch": 0.25,
|
1107 |
-
"learning_rate": 4.525000000000001e-06,
|
1108 |
-
"loss": 3.2969,
|
1109 |
-
"step": 910
|
1110 |
-
},
|
1111 |
-
{
|
1112 |
-
"epoch": 0.26,
|
1113 |
-
"learning_rate": 4.5500000000000005e-06,
|
1114 |
-
"loss": 3.2062,
|
1115 |
-
"step": 915
|
1116 |
-
},
|
1117 |
-
{
|
1118 |
-
"epoch": 0.26,
|
1119 |
-
"learning_rate": 4.575e-06,
|
1120 |
-
"loss": 3.1091,
|
1121 |
-
"step": 920
|
1122 |
-
},
|
1123 |
-
{
|
1124 |
-
"epoch": 0.26,
|
1125 |
-
"learning_rate": 4.600000000000001e-06,
|
1126 |
-
"loss": 3.288,
|
1127 |
-
"step": 925
|
1128 |
-
},
|
1129 |
-
{
|
1130 |
-
"epoch": 0.26,
|
1131 |
-
"learning_rate": 4.625000000000001e-06,
|
1132 |
-
"loss": 2.8966,
|
1133 |
-
"step": 930
|
1134 |
-
},
|
1135 |
-
{
|
1136 |
-
"epoch": 0.26,
|
1137 |
-
"learning_rate": 4.65e-06,
|
1138 |
-
"loss": 2.9091,
|
1139 |
-
"step": 935
|
1140 |
-
},
|
1141 |
-
{
|
1142 |
-
"epoch": 0.26,
|
1143 |
-
"learning_rate": 4.675000000000001e-06,
|
1144 |
-
"loss": 3.212,
|
1145 |
-
"step": 940
|
1146 |
-
},
|
1147 |
-
{
|
1148 |
-
"epoch": 0.26,
|
1149 |
-
"learning_rate": 4.7e-06,
|
1150 |
-
"loss": 2.8426,
|
1151 |
-
"step": 945
|
1152 |
-
},
|
1153 |
-
{
|
1154 |
-
"epoch": 0.27,
|
1155 |
-
"learning_rate": 4.7250000000000005e-06,
|
1156 |
-
"loss": 3.1961,
|
1157 |
-
"step": 950
|
1158 |
-
},
|
1159 |
-
{
|
1160 |
-
"epoch": 0.27,
|
1161 |
-
"learning_rate": 4.75e-06,
|
1162 |
-
"loss": 2.9483,
|
1163 |
-
"step": 955
|
1164 |
-
},
|
1165 |
-
{
|
1166 |
-
"epoch": 0.27,
|
1167 |
-
"learning_rate": 4.775e-06,
|
1168 |
-
"loss": 2.9817,
|
1169 |
-
"step": 960
|
1170 |
-
},
|
1171 |
-
{
|
1172 |
-
"epoch": 0.27,
|
1173 |
-
"learning_rate": 4.800000000000001e-06,
|
1174 |
-
"loss": 3.0139,
|
1175 |
-
"step": 965
|
1176 |
-
},
|
1177 |
-
{
|
1178 |
-
"epoch": 0.27,
|
1179 |
-
"learning_rate": 4.825e-06,
|
1180 |
-
"loss": 2.8438,
|
1181 |
-
"step": 970
|
1182 |
-
},
|
1183 |
-
{
|
1184 |
-
"epoch": 0.27,
|
1185 |
-
"learning_rate": 4.85e-06,
|
1186 |
-
"loss": 3.008,
|
1187 |
-
"step": 975
|
1188 |
-
},
|
1189 |
-
{
|
1190 |
-
"epoch": 0.27,
|
1191 |
-
"learning_rate": 4.875e-06,
|
1192 |
-
"loss": 2.9319,
|
1193 |
-
"step": 980
|
1194 |
-
},
|
1195 |
-
{
|
1196 |
-
"epoch": 0.28,
|
1197 |
-
"learning_rate": 4.9000000000000005e-06,
|
1198 |
-
"loss": 2.8704,
|
1199 |
-
"step": 985
|
1200 |
-
},
|
1201 |
-
{
|
1202 |
-
"epoch": 0.28,
|
1203 |
-
"learning_rate": 4.925e-06,
|
1204 |
-
"loss": 3.0068,
|
1205 |
-
"step": 990
|
1206 |
-
},
|
1207 |
-
{
|
1208 |
-
"epoch": 0.28,
|
1209 |
-
"learning_rate": 4.95e-06,
|
1210 |
-
"loss": 2.8494,
|
1211 |
-
"step": 995
|
1212 |
-
},
|
1213 |
-
{
|
1214 |
-
"epoch": 0.28,
|
1215 |
-
"learning_rate": 4.975000000000001e-06,
|
1216 |
-
"loss": 2.8652,
|
1217 |
-
"step": 1000
|
1218 |
-
},
|
1219 |
-
{
|
1220 |
-
"epoch": 0.28,
|
1221 |
-
"eval_loss": 2.881692886352539,
|
1222 |
-
"eval_runtime": 141.6494,
|
1223 |
-
"eval_samples_per_second": 8.429,
|
1224 |
-
"eval_steps_per_second": 1.059,
|
1225 |
-
"eval_wer": 1.0,
|
1226 |
-
"step": 1000
|
1227 |
-
},
|
1228 |
-
{
|
1229 |
-
"epoch": 0.28,
|
1230 |
-
"learning_rate": 5e-06,
|
1231 |
-
"loss": 2.9299,
|
1232 |
-
"step": 1005
|
1233 |
-
},
|
1234 |
-
{
|
1235 |
-
"epoch": 0.28,
|
1236 |
-
"learning_rate": 4.998122559327126e-06,
|
1237 |
-
"loss": 2.855,
|
1238 |
-
"step": 1010
|
1239 |
-
},
|
1240 |
-
{
|
1241 |
-
"epoch": 0.28,
|
1242 |
-
"learning_rate": 4.996245118654251e-06,
|
1243 |
-
"loss": 2.8683,
|
1244 |
-
"step": 1015
|
1245 |
-
},
|
1246 |
-
{
|
1247 |
-
"epoch": 0.28,
|
1248 |
-
"learning_rate": 4.994367677981376e-06,
|
1249 |
-
"loss": 2.8563,
|
1250 |
-
"step": 1020
|
1251 |
-
},
|
1252 |
-
{
|
1253 |
-
"epoch": 0.29,
|
1254 |
-
"learning_rate": 4.992490237308501e-06,
|
1255 |
-
"loss": 2.9462,
|
1256 |
-
"step": 1025
|
1257 |
-
},
|
1258 |
-
{
|
1259 |
-
"epoch": 0.29,
|
1260 |
-
"learning_rate": 4.990612796635627e-06,
|
1261 |
-
"loss": 2.8713,
|
1262 |
-
"step": 1030
|
1263 |
-
},
|
1264 |
-
{
|
1265 |
-
"epoch": 0.29,
|
1266 |
-
"learning_rate": 4.988735355962752e-06,
|
1267 |
-
"loss": 2.9464,
|
1268 |
-
"step": 1035
|
1269 |
-
},
|
1270 |
-
{
|
1271 |
-
"epoch": 0.29,
|
1272 |
-
"learning_rate": 4.986857915289877e-06,
|
1273 |
-
"loss": 2.8587,
|
1274 |
-
"step": 1040
|
1275 |
-
},
|
1276 |
-
{
|
1277 |
-
"epoch": 0.29,
|
1278 |
-
"learning_rate": 4.984980474617002e-06,
|
1279 |
-
"loss": 2.9491,
|
1280 |
-
"step": 1045
|
1281 |
-
},
|
1282 |
-
{
|
1283 |
-
"epoch": 0.29,
|
1284 |
-
"learning_rate": 4.9831030339441276e-06,
|
1285 |
-
"loss": 2.8847,
|
1286 |
-
"step": 1050
|
1287 |
-
},
|
1288 |
-
{
|
1289 |
-
"epoch": 0.29,
|
1290 |
-
"learning_rate": 4.981225593271253e-06,
|
1291 |
-
"loss": 2.9443,
|
1292 |
-
"step": 1055
|
1293 |
-
},
|
1294 |
-
{
|
1295 |
-
"epoch": 0.3,
|
1296 |
-
"learning_rate": 4.979348152598378e-06,
|
1297 |
-
"loss": 2.9,
|
1298 |
-
"step": 1060
|
1299 |
-
},
|
1300 |
-
{
|
1301 |
-
"epoch": 0.3,
|
1302 |
-
"learning_rate": 4.977470711925503e-06,
|
1303 |
-
"loss": 2.8448,
|
1304 |
-
"step": 1065
|
1305 |
-
},
|
1306 |
-
{
|
1307 |
-
"epoch": 0.3,
|
1308 |
-
"learning_rate": 4.9755932712526285e-06,
|
1309 |
-
"loss": 2.862,
|
1310 |
-
"step": 1070
|
1311 |
-
},
|
1312 |
-
{
|
1313 |
-
"epoch": 0.3,
|
1314 |
-
"learning_rate": 4.973715830579754e-06,
|
1315 |
-
"loss": 2.8782,
|
1316 |
-
"step": 1075
|
1317 |
-
},
|
1318 |
-
{
|
1319 |
-
"epoch": 0.3,
|
1320 |
-
"learning_rate": 4.97183838990688e-06,
|
1321 |
-
"loss": 2.866,
|
1322 |
-
"step": 1080
|
1323 |
-
},
|
1324 |
-
{
|
1325 |
-
"epoch": 0.3,
|
1326 |
-
"learning_rate": 4.969960949234004e-06,
|
1327 |
-
"loss": 2.9187,
|
1328 |
-
"step": 1085
|
1329 |
-
},
|
1330 |
-
{
|
1331 |
-
"epoch": 0.3,
|
1332 |
-
"learning_rate": 4.96808350856113e-06,
|
1333 |
-
"loss": 2.8627,
|
1334 |
-
"step": 1090
|
1335 |
-
},
|
1336 |
-
{
|
1337 |
-
"epoch": 0.31,
|
1338 |
-
"learning_rate": 4.966206067888255e-06,
|
1339 |
-
"loss": 2.8636,
|
1340 |
-
"step": 1095
|
1341 |
-
},
|
1342 |
-
{
|
1343 |
-
"epoch": 0.31,
|
1344 |
-
"learning_rate": 4.964328627215381e-06,
|
1345 |
-
"loss": 2.9201,
|
1346 |
-
"step": 1100
|
1347 |
-
},
|
1348 |
-
{
|
1349 |
-
"epoch": 0.31,
|
1350 |
-
"learning_rate": 4.962451186542506e-06,
|
1351 |
-
"loss": 2.9078,
|
1352 |
-
"step": 1105
|
1353 |
-
},
|
1354 |
-
{
|
1355 |
-
"epoch": 0.31,
|
1356 |
-
"learning_rate": 4.9605737458696305e-06,
|
1357 |
-
"loss": 2.8853,
|
1358 |
-
"step": 1110
|
1359 |
-
},
|
1360 |
-
{
|
1361 |
-
"epoch": 0.31,
|
1362 |
-
"learning_rate": 4.9586963051967565e-06,
|
1363 |
-
"loss": 2.8612,
|
1364 |
-
"step": 1115
|
1365 |
-
},
|
1366 |
-
{
|
1367 |
-
"epoch": 0.31,
|
1368 |
-
"learning_rate": 4.956818864523881e-06,
|
1369 |
-
"loss": 2.8607,
|
1370 |
-
"step": 1120
|
1371 |
-
},
|
1372 |
-
{
|
1373 |
-
"epoch": 0.31,
|
1374 |
-
"learning_rate": 4.954941423851007e-06,
|
1375 |
-
"loss": 2.845,
|
1376 |
-
"step": 1125
|
1377 |
-
},
|
1378 |
-
{
|
1379 |
-
"epoch": 0.32,
|
1380 |
-
"learning_rate": 4.953063983178132e-06,
|
1381 |
-
"loss": 2.8778,
|
1382 |
-
"step": 1130
|
1383 |
-
},
|
1384 |
-
{
|
1385 |
-
"epoch": 0.32,
|
1386 |
-
"learning_rate": 4.9511865425052575e-06,
|
1387 |
-
"loss": 3.1876,
|
1388 |
-
"step": 1135
|
1389 |
-
},
|
1390 |
-
{
|
1391 |
-
"epoch": 0.32,
|
1392 |
-
"learning_rate": 4.949309101832383e-06,
|
1393 |
-
"loss": 2.8603,
|
1394 |
-
"step": 1140
|
1395 |
-
},
|
1396 |
-
{
|
1397 |
-
"epoch": 0.32,
|
1398 |
-
"learning_rate": 4.947431661159508e-06,
|
1399 |
-
"loss": 2.8403,
|
1400 |
-
"step": 1145
|
1401 |
-
},
|
1402 |
-
{
|
1403 |
-
"epoch": 0.32,
|
1404 |
-
"learning_rate": 4.945554220486633e-06,
|
1405 |
-
"loss": 2.8791,
|
1406 |
-
"step": 1150
|
1407 |
-
},
|
1408 |
-
{
|
1409 |
-
"epoch": 0.32,
|
1410 |
-
"learning_rate": 4.9436767798137585e-06,
|
1411 |
-
"loss": 2.8604,
|
1412 |
-
"step": 1155
|
1413 |
-
},
|
1414 |
-
{
|
1415 |
-
"epoch": 0.32,
|
1416 |
-
"learning_rate": 4.941799339140884e-06,
|
1417 |
-
"loss": 2.8595,
|
1418 |
-
"step": 1160
|
1419 |
-
},
|
1420 |
-
{
|
1421 |
-
"epoch": 0.33,
|
1422 |
-
"learning_rate": 4.939921898468009e-06,
|
1423 |
-
"loss": 2.9011,
|
1424 |
-
"step": 1165
|
1425 |
-
},
|
1426 |
-
{
|
1427 |
-
"epoch": 0.33,
|
1428 |
-
"learning_rate": 4.938044457795134e-06,
|
1429 |
-
"loss": 2.86,
|
1430 |
-
"step": 1170
|
1431 |
-
},
|
1432 |
-
{
|
1433 |
-
"epoch": 0.33,
|
1434 |
-
"learning_rate": 4.936167017122259e-06,
|
1435 |
-
"loss": 2.8583,
|
1436 |
-
"step": 1175
|
1437 |
-
},
|
1438 |
-
{
|
1439 |
-
"epoch": 0.33,
|
1440 |
-
"learning_rate": 4.934289576449385e-06,
|
1441 |
-
"loss": 2.9274,
|
1442 |
-
"step": 1180
|
1443 |
-
},
|
1444 |
-
{
|
1445 |
-
"epoch": 0.33,
|
1446 |
-
"learning_rate": 4.93241213577651e-06,
|
1447 |
-
"loss": 2.9472,
|
1448 |
-
"step": 1185
|
1449 |
-
},
|
1450 |
-
{
|
1451 |
-
"epoch": 0.33,
|
1452 |
-
"learning_rate": 4.930534695103635e-06,
|
1453 |
-
"loss": 3.4067,
|
1454 |
-
"step": 1190
|
1455 |
-
},
|
1456 |
-
{
|
1457 |
-
"epoch": 0.33,
|
1458 |
-
"learning_rate": 4.92865725443076e-06,
|
1459 |
-
"loss": 2.9363,
|
1460 |
-
"step": 1195
|
1461 |
-
},
|
1462 |
-
{
|
1463 |
-
"epoch": 0.34,
|
1464 |
-
"learning_rate": 4.926779813757886e-06,
|
1465 |
-
"loss": 2.8608,
|
1466 |
-
"step": 1200
|
1467 |
-
},
|
1468 |
-
{
|
1469 |
-
"epoch": 0.34,
|
1470 |
-
"learning_rate": 4.924902373085011e-06,
|
1471 |
-
"loss": 3.356,
|
1472 |
-
"step": 1205
|
1473 |
-
},
|
1474 |
-
{
|
1475 |
-
"epoch": 0.34,
|
1476 |
-
"learning_rate": 4.923024932412136e-06,
|
1477 |
-
"loss": 2.8678,
|
1478 |
-
"step": 1210
|
1479 |
-
},
|
1480 |
-
{
|
1481 |
-
"epoch": 0.34,
|
1482 |
-
"learning_rate": 4.921147491739261e-06,
|
1483 |
-
"loss": 2.8664,
|
1484 |
-
"step": 1215
|
1485 |
-
},
|
1486 |
-
{
|
1487 |
-
"epoch": 0.34,
|
1488 |
-
"learning_rate": 4.919270051066387e-06,
|
1489 |
-
"loss": 2.8897,
|
1490 |
-
"step": 1220
|
1491 |
-
},
|
1492 |
-
{
|
1493 |
-
"epoch": 0.34,
|
1494 |
-
"learning_rate": 4.917392610393512e-06,
|
1495 |
-
"loss": 2.8783,
|
1496 |
-
"step": 1225
|
1497 |
-
},
|
1498 |
-
{
|
1499 |
-
"epoch": 0.34,
|
1500 |
-
"learning_rate": 4.915515169720637e-06,
|
1501 |
-
"loss": 2.8886,
|
1502 |
-
"step": 1230
|
1503 |
-
},
|
1504 |
-
{
|
1505 |
-
"epoch": 0.35,
|
1506 |
-
"learning_rate": 4.913637729047762e-06,
|
1507 |
-
"loss": 2.8534,
|
1508 |
-
"step": 1235
|
1509 |
-
},
|
1510 |
-
{
|
1511 |
-
"epoch": 0.35,
|
1512 |
-
"learning_rate": 4.9117602883748875e-06,
|
1513 |
-
"loss": 2.8322,
|
1514 |
-
"step": 1240
|
1515 |
-
},
|
1516 |
-
{
|
1517 |
-
"epoch": 0.35,
|
1518 |
-
"learning_rate": 4.909882847702014e-06,
|
1519 |
-
"loss": 2.8148,
|
1520 |
-
"step": 1245
|
1521 |
-
},
|
1522 |
-
{
|
1523 |
-
"epoch": 0.35,
|
1524 |
-
"learning_rate": 4.908005407029138e-06,
|
1525 |
-
"loss": 2.832,
|
1526 |
-
"step": 1250
|
1527 |
-
},
|
1528 |
-
{
|
1529 |
-
"epoch": 0.35,
|
1530 |
-
"learning_rate": 4.906127966356263e-06,
|
1531 |
-
"loss": 2.8783,
|
1532 |
-
"step": 1255
|
1533 |
-
},
|
1534 |
-
{
|
1535 |
-
"epoch": 0.35,
|
1536 |
-
"learning_rate": 4.9042505256833885e-06,
|
1537 |
-
"loss": 2.8255,
|
1538 |
-
"step": 1260
|
1539 |
-
},
|
1540 |
-
{
|
1541 |
-
"epoch": 0.35,
|
1542 |
-
"learning_rate": 4.902373085010514e-06,
|
1543 |
-
"loss": 2.8893,
|
1544 |
-
"step": 1265
|
1545 |
-
},
|
1546 |
-
{
|
1547 |
-
"epoch": 0.35,
|
1548 |
-
"learning_rate": 4.90049564433764e-06,
|
1549 |
-
"loss": 2.9132,
|
1550 |
-
"step": 1270
|
1551 |
-
},
|
1552 |
-
{
|
1553 |
-
"epoch": 0.36,
|
1554 |
-
"learning_rate": 4.898618203664764e-06,
|
1555 |
-
"loss": 2.8366,
|
1556 |
-
"step": 1275
|
1557 |
-
},
|
1558 |
-
{
|
1559 |
-
"epoch": 0.36,
|
1560 |
-
"learning_rate": 4.89674076299189e-06,
|
1561 |
-
"loss": 2.8717,
|
1562 |
-
"step": 1280
|
1563 |
-
},
|
1564 |
-
{
|
1565 |
-
"epoch": 0.36,
|
1566 |
-
"learning_rate": 4.894863322319015e-06,
|
1567 |
-
"loss": 3.0332,
|
1568 |
-
"step": 1285
|
1569 |
-
},
|
1570 |
-
{
|
1571 |
-
"epoch": 0.36,
|
1572 |
-
"learning_rate": 4.892985881646141e-06,
|
1573 |
-
"loss": 2.8806,
|
1574 |
-
"step": 1290
|
1575 |
-
},
|
1576 |
-
{
|
1577 |
-
"epoch": 0.36,
|
1578 |
-
"learning_rate": 4.891108440973266e-06,
|
1579 |
-
"loss": 2.8377,
|
1580 |
-
"step": 1295
|
1581 |
-
},
|
1582 |
-
{
|
1583 |
-
"epoch": 0.36,
|
1584 |
-
"learning_rate": 4.88923100030039e-06,
|
1585 |
-
"loss": 2.8136,
|
1586 |
-
"step": 1300
|
1587 |
-
},
|
1588 |
-
{
|
1589 |
-
"epoch": 0.36,
|
1590 |
-
"learning_rate": 4.8873535596275165e-06,
|
1591 |
-
"loss": 2.9869,
|
1592 |
-
"step": 1305
|
1593 |
-
},
|
1594 |
-
{
|
1595 |
-
"epoch": 0.37,
|
1596 |
-
"learning_rate": 4.885476118954641e-06,
|
1597 |
-
"loss": 2.8812,
|
1598 |
-
"step": 1310
|
1599 |
-
},
|
1600 |
-
{
|
1601 |
-
"epoch": 0.37,
|
1602 |
-
"learning_rate": 4.883598678281767e-06,
|
1603 |
-
"loss": 2.83,
|
1604 |
-
"step": 1315
|
1605 |
-
},
|
1606 |
-
{
|
1607 |
-
"epoch": 0.37,
|
1608 |
-
"learning_rate": 4.881721237608892e-06,
|
1609 |
-
"loss": 2.8324,
|
1610 |
-
"step": 1320
|
1611 |
-
},
|
1612 |
-
{
|
1613 |
-
"epoch": 0.37,
|
1614 |
-
"learning_rate": 4.8798437969360175e-06,
|
1615 |
-
"loss": 2.829,
|
1616 |
-
"step": 1325
|
1617 |
-
},
|
1618 |
-
{
|
1619 |
-
"epoch": 0.37,
|
1620 |
-
"learning_rate": 4.877966356263143e-06,
|
1621 |
-
"loss": 2.834,
|
1622 |
-
"step": 1330
|
1623 |
-
},
|
1624 |
-
{
|
1625 |
-
"epoch": 0.37,
|
1626 |
-
"learning_rate": 4.876088915590268e-06,
|
1627 |
-
"loss": 3.1544,
|
1628 |
-
"step": 1335
|
1629 |
-
},
|
1630 |
-
{
|
1631 |
-
"epoch": 0.37,
|
1632 |
-
"learning_rate": 4.874211474917393e-06,
|
1633 |
-
"loss": 2.8126,
|
1634 |
-
"step": 1340
|
1635 |
-
},
|
1636 |
-
{
|
1637 |
-
"epoch": 0.38,
|
1638 |
-
"learning_rate": 4.8723340342445184e-06,
|
1639 |
-
"loss": 2.8915,
|
1640 |
-
"step": 1345
|
1641 |
-
},
|
1642 |
-
{
|
1643 |
-
"epoch": 0.38,
|
1644 |
-
"learning_rate": 4.870456593571644e-06,
|
1645 |
-
"loss": 3.0785,
|
1646 |
-
"step": 1350
|
1647 |
-
},
|
1648 |
-
{
|
1649 |
-
"epoch": 0.38,
|
1650 |
-
"learning_rate": 4.868579152898769e-06,
|
1651 |
-
"loss": 2.8628,
|
1652 |
-
"step": 1355
|
1653 |
-
},
|
1654 |
-
{
|
1655 |
-
"epoch": 0.38,
|
1656 |
-
"learning_rate": 4.866701712225894e-06,
|
1657 |
-
"loss": 2.8263,
|
1658 |
-
"step": 1360
|
1659 |
-
},
|
1660 |
-
{
|
1661 |
-
"epoch": 0.38,
|
1662 |
-
"learning_rate": 4.864824271553019e-06,
|
1663 |
-
"loss": 2.8055,
|
1664 |
-
"step": 1365
|
1665 |
-
},
|
1666 |
-
{
|
1667 |
-
"epoch": 0.38,
|
1668 |
-
"learning_rate": 4.862946830880145e-06,
|
1669 |
-
"loss": 2.837,
|
1670 |
-
"step": 1370
|
1671 |
-
},
|
1672 |
-
{
|
1673 |
-
"epoch": 0.38,
|
1674 |
-
"learning_rate": 4.86106939020727e-06,
|
1675 |
-
"loss": 2.8109,
|
1676 |
-
"step": 1375
|
1677 |
-
},
|
1678 |
-
{
|
1679 |
-
"epoch": 0.39,
|
1680 |
-
"learning_rate": 4.859191949534395e-06,
|
1681 |
-
"loss": 2.8313,
|
1682 |
-
"step": 1380
|
1683 |
-
},
|
1684 |
-
{
|
1685 |
-
"epoch": 0.39,
|
1686 |
-
"learning_rate": 4.85731450886152e-06,
|
1687 |
-
"loss": 2.8085,
|
1688 |
-
"step": 1385
|
1689 |
-
},
|
1690 |
-
{
|
1691 |
-
"epoch": 0.39,
|
1692 |
-
"learning_rate": 4.855437068188646e-06,
|
1693 |
-
"loss": 2.9006,
|
1694 |
-
"step": 1390
|
1695 |
-
},
|
1696 |
-
{
|
1697 |
-
"epoch": 0.39,
|
1698 |
-
"learning_rate": 4.853559627515771e-06,
|
1699 |
-
"loss": 2.76,
|
1700 |
-
"step": 1395
|
1701 |
-
},
|
1702 |
-
{
|
1703 |
-
"epoch": 0.39,
|
1704 |
-
"learning_rate": 4.851682186842896e-06,
|
1705 |
-
"loss": 2.7712,
|
1706 |
-
"step": 1400
|
1707 |
-
},
|
1708 |
-
{
|
1709 |
-
"epoch": 0.39,
|
1710 |
-
"learning_rate": 4.849804746170021e-06,
|
1711 |
-
"loss": 2.8082,
|
1712 |
-
"step": 1405
|
1713 |
-
},
|
1714 |
-
{
|
1715 |
-
"epoch": 0.39,
|
1716 |
-
"learning_rate": 4.8479273054971466e-06,
|
1717 |
-
"loss": 2.9101,
|
1718 |
-
"step": 1410
|
1719 |
-
},
|
1720 |
-
{
|
1721 |
-
"epoch": 0.4,
|
1722 |
-
"learning_rate": 4.846049864824272e-06,
|
1723 |
-
"loss": 2.8121,
|
1724 |
-
"step": 1415
|
1725 |
-
},
|
1726 |
-
{
|
1727 |
-
"epoch": 0.4,
|
1728 |
-
"learning_rate": 4.844172424151397e-06,
|
1729 |
-
"loss": 2.7971,
|
1730 |
-
"step": 1420
|
1731 |
-
},
|
1732 |
-
{
|
1733 |
-
"epoch": 0.4,
|
1734 |
-
"learning_rate": 4.842294983478522e-06,
|
1735 |
-
"loss": 2.8376,
|
1736 |
-
"step": 1425
|
1737 |
-
},
|
1738 |
-
{
|
1739 |
-
"epoch": 0.4,
|
1740 |
-
"learning_rate": 4.8404175428056475e-06,
|
1741 |
-
"loss": 2.7967,
|
1742 |
-
"step": 1430
|
1743 |
-
},
|
1744 |
-
{
|
1745 |
-
"epoch": 0.4,
|
1746 |
-
"learning_rate": 4.838540102132774e-06,
|
1747 |
-
"loss": 2.8066,
|
1748 |
-
"step": 1435
|
1749 |
-
},
|
1750 |
-
{
|
1751 |
-
"epoch": 0.4,
|
1752 |
-
"learning_rate": 4.836662661459898e-06,
|
1753 |
-
"loss": 2.7832,
|
1754 |
-
"step": 1440
|
1755 |
-
},
|
1756 |
-
{
|
1757 |
-
"epoch": 0.4,
|
1758 |
-
"learning_rate": 4.834785220787023e-06,
|
1759 |
-
"loss": 2.7981,
|
1760 |
-
"step": 1445
|
1761 |
-
},
|
1762 |
-
{
|
1763 |
-
"epoch": 0.41,
|
1764 |
-
"learning_rate": 4.8329077801141485e-06,
|
1765 |
-
"loss": 2.743,
|
1766 |
-
"step": 1450
|
1767 |
-
},
|
1768 |
-
{
|
1769 |
-
"epoch": 0.41,
|
1770 |
-
"learning_rate": 4.831030339441274e-06,
|
1771 |
-
"loss": 2.8466,
|
1772 |
-
"step": 1455
|
1773 |
-
},
|
1774 |
-
{
|
1775 |
-
"epoch": 0.41,
|
1776 |
-
"learning_rate": 4.8291528987684e-06,
|
1777 |
-
"loss": 2.8222,
|
1778 |
-
"step": 1460
|
1779 |
-
},
|
1780 |
-
{
|
1781 |
-
"epoch": 0.41,
|
1782 |
-
"learning_rate": 4.827275458095524e-06,
|
1783 |
-
"loss": 2.802,
|
1784 |
-
"step": 1465
|
1785 |
-
},
|
1786 |
-
{
|
1787 |
-
"epoch": 0.41,
|
1788 |
-
"learning_rate": 4.82539801742265e-06,
|
1789 |
-
"loss": 2.6981,
|
1790 |
-
"step": 1470
|
1791 |
-
},
|
1792 |
-
{
|
1793 |
-
"epoch": 0.41,
|
1794 |
-
"learning_rate": 4.823520576749775e-06,
|
1795 |
-
"loss": 2.7656,
|
1796 |
-
"step": 1475
|
1797 |
-
},
|
1798 |
-
{
|
1799 |
-
"epoch": 0.41,
|
1800 |
-
"learning_rate": 4.821643136076901e-06,
|
1801 |
-
"loss": 2.6938,
|
1802 |
-
"step": 1480
|
1803 |
-
},
|
1804 |
-
{
|
1805 |
-
"epoch": 0.41,
|
1806 |
-
"learning_rate": 4.819765695404026e-06,
|
1807 |
-
"loss": 2.7163,
|
1808 |
-
"step": 1485
|
1809 |
-
},
|
1810 |
-
{
|
1811 |
-
"epoch": 0.42,
|
1812 |
-
"learning_rate": 4.81788825473115e-06,
|
1813 |
-
"loss": 2.7057,
|
1814 |
-
"step": 1490
|
1815 |
-
},
|
1816 |
-
{
|
1817 |
-
"epoch": 0.42,
|
1818 |
-
"learning_rate": 4.8160108140582765e-06,
|
1819 |
-
"loss": 2.7461,
|
1820 |
-
"step": 1495
|
1821 |
-
},
|
1822 |
-
{
|
1823 |
-
"epoch": 0.42,
|
1824 |
-
"learning_rate": 4.814133373385401e-06,
|
1825 |
-
"loss": 2.7789,
|
1826 |
-
"step": 1500
|
1827 |
-
},
|
1828 |
-
{
|
1829 |
-
"epoch": 0.42,
|
1830 |
-
"eval_loss": 2.633471965789795,
|
1831 |
-
"eval_runtime": 154.8101,
|
1832 |
-
"eval_samples_per_second": 7.713,
|
1833 |
-
"eval_steps_per_second": 0.969,
|
1834 |
-
"eval_wer": 1.0,
|
1835 |
-
"step": 1500
|
1836 |
-
},
|
1837 |
-
{
|
1838 |
-
"epoch": 0.42,
|
1839 |
-
"learning_rate": 4.812255932712527e-06,
|
1840 |
-
"loss": 2.7559,
|
1841 |
-
"step": 1505
|
1842 |
-
},
|
1843 |
-
{
|
1844 |
-
"epoch": 0.42,
|
1845 |
-
"learning_rate": 4.810378492039652e-06,
|
1846 |
-
"loss": 3.0039,
|
1847 |
-
"step": 1510
|
1848 |
-
},
|
1849 |
-
{
|
1850 |
-
"epoch": 0.42,
|
1851 |
-
"learning_rate": 4.8085010513667774e-06,
|
1852 |
-
"loss": 2.6165,
|
1853 |
-
"step": 1515
|
1854 |
-
},
|
1855 |
-
{
|
1856 |
-
"epoch": 0.42,
|
1857 |
-
"learning_rate": 4.806623610693903e-06,
|
1858 |
-
"loss": 3.5613,
|
1859 |
-
"step": 1520
|
1860 |
-
},
|
1861 |
-
{
|
1862 |
-
"epoch": 0.43,
|
1863 |
-
"learning_rate": 4.804746170021028e-06,
|
1864 |
-
"loss": 2.5695,
|
1865 |
-
"step": 1525
|
1866 |
-
},
|
1867 |
-
{
|
1868 |
-
"epoch": 0.43,
|
1869 |
-
"learning_rate": 4.802868729348153e-06,
|
1870 |
-
"loss": 2.6025,
|
1871 |
-
"step": 1530
|
1872 |
-
},
|
1873 |
-
{
|
1874 |
-
"epoch": 0.43,
|
1875 |
-
"learning_rate": 4.800991288675278e-06,
|
1876 |
-
"loss": 2.6736,
|
1877 |
-
"step": 1535
|
1878 |
-
},
|
1879 |
-
{
|
1880 |
-
"epoch": 0.43,
|
1881 |
-
"learning_rate": 4.799113848002404e-06,
|
1882 |
-
"loss": 2.5119,
|
1883 |
-
"step": 1540
|
1884 |
-
},
|
1885 |
-
{
|
1886 |
-
"epoch": 0.43,
|
1887 |
-
"learning_rate": 4.797236407329529e-06,
|
1888 |
-
"loss": 2.7166,
|
1889 |
-
"step": 1545
|
1890 |
-
},
|
1891 |
-
{
|
1892 |
-
"epoch": 0.43,
|
1893 |
-
"learning_rate": 4.795358966656654e-06,
|
1894 |
-
"loss": 2.5151,
|
1895 |
-
"step": 1550
|
1896 |
-
},
|
1897 |
-
{
|
1898 |
-
"epoch": 0.43,
|
1899 |
-
"learning_rate": 4.793481525983779e-06,
|
1900 |
-
"loss": 2.5859,
|
1901 |
-
"step": 1555
|
1902 |
-
},
|
1903 |
-
{
|
1904 |
-
"epoch": 0.44,
|
1905 |
-
"learning_rate": 4.791604085310905e-06,
|
1906 |
-
"loss": 2.4956,
|
1907 |
-
"step": 1560
|
1908 |
-
},
|
1909 |
-
{
|
1910 |
-
"epoch": 0.44,
|
1911 |
-
"learning_rate": 4.78972664463803e-06,
|
1912 |
-
"loss": 2.638,
|
1913 |
-
"step": 1565
|
1914 |
-
},
|
1915 |
-
{
|
1916 |
-
"epoch": 0.44,
|
1917 |
-
"learning_rate": 4.787849203965155e-06,
|
1918 |
-
"loss": 2.5352,
|
1919 |
-
"step": 1570
|
1920 |
-
},
|
1921 |
-
{
|
1922 |
-
"epoch": 0.44,
|
1923 |
-
"learning_rate": 4.78597176329228e-06,
|
1924 |
-
"loss": 2.4668,
|
1925 |
-
"step": 1575
|
1926 |
-
},
|
1927 |
-
{
|
1928 |
-
"epoch": 0.44,
|
1929 |
-
"learning_rate": 4.7840943226194056e-06,
|
1930 |
-
"loss": 2.5222,
|
1931 |
-
"step": 1580
|
1932 |
-
},
|
1933 |
-
{
|
1934 |
-
"epoch": 0.44,
|
1935 |
-
"learning_rate": 4.782216881946531e-06,
|
1936 |
-
"loss": 2.6859,
|
1937 |
-
"step": 1585
|
1938 |
-
},
|
1939 |
-
{
|
1940 |
-
"epoch": 0.44,
|
1941 |
-
"learning_rate": 4.780339441273656e-06,
|
1942 |
-
"loss": 2.4609,
|
1943 |
-
"step": 1590
|
1944 |
-
},
|
1945 |
-
{
|
1946 |
-
"epoch": 0.45,
|
1947 |
-
"learning_rate": 4.778462000600781e-06,
|
1948 |
-
"loss": 2.4346,
|
1949 |
-
"step": 1595
|
1950 |
-
},
|
1951 |
-
{
|
1952 |
-
"epoch": 0.45,
|
1953 |
-
"learning_rate": 4.7765845599279065e-06,
|
1954 |
-
"loss": 2.4685,
|
1955 |
-
"step": 1600
|
1956 |
-
},
|
1957 |
-
{
|
1958 |
-
"epoch": 0.45,
|
1959 |
-
"learning_rate": 4.774707119255032e-06,
|
1960 |
-
"loss": 2.4568,
|
1961 |
-
"step": 1605
|
1962 |
-
},
|
1963 |
-
{
|
1964 |
-
"epoch": 0.45,
|
1965 |
-
"learning_rate": 4.772829678582157e-06,
|
1966 |
-
"loss": 2.2941,
|
1967 |
-
"step": 1610
|
1968 |
-
},
|
1969 |
-
{
|
1970 |
-
"epoch": 0.45,
|
1971 |
-
"learning_rate": 4.770952237909282e-06,
|
1972 |
-
"loss": 2.2794,
|
1973 |
-
"step": 1615
|
1974 |
-
},
|
1975 |
-
{
|
1976 |
-
"epoch": 0.45,
|
1977 |
-
"learning_rate": 4.7690747972364075e-06,
|
1978 |
-
"loss": 2.4405,
|
1979 |
-
"step": 1620
|
1980 |
-
},
|
1981 |
-
{
|
1982 |
-
"epoch": 0.45,
|
1983 |
-
"learning_rate": 4.7671973565635336e-06,
|
1984 |
-
"loss": 2.3013,
|
1985 |
-
"step": 1625
|
1986 |
-
},
|
1987 |
-
{
|
1988 |
-
"epoch": 0.46,
|
1989 |
-
"learning_rate": 4.765319915890658e-06,
|
1990 |
-
"loss": 2.2069,
|
1991 |
-
"step": 1630
|
1992 |
-
},
|
1993 |
-
{
|
1994 |
-
"epoch": 0.46,
|
1995 |
-
"learning_rate": 4.763442475217783e-06,
|
1996 |
-
"loss": 2.1934,
|
1997 |
-
"step": 1635
|
1998 |
-
},
|
1999 |
-
{
|
2000 |
-
"epoch": 0.46,
|
2001 |
-
"learning_rate": 4.7615650345449084e-06,
|
2002 |
-
"loss": 2.1879,
|
2003 |
-
"step": 1640
|
2004 |
-
},
|
2005 |
-
{
|
2006 |
-
"epoch": 0.46,
|
2007 |
-
"learning_rate": 4.759687593872034e-06,
|
2008 |
-
"loss": 2.1863,
|
2009 |
-
"step": 1645
|
2010 |
-
},
|
2011 |
-
{
|
2012 |
-
"epoch": 0.46,
|
2013 |
-
"learning_rate": 4.75781015319916e-06,
|
2014 |
-
"loss": 2.2998,
|
2015 |
-
"step": 1650
|
2016 |
-
},
|
2017 |
-
{
|
2018 |
-
"epoch": 0.46,
|
2019 |
-
"learning_rate": 4.755932712526284e-06,
|
2020 |
-
"loss": 2.3798,
|
2021 |
-
"step": 1655
|
2022 |
-
},
|
2023 |
-
{
|
2024 |
-
"epoch": 0.46,
|
2025 |
-
"learning_rate": 4.75405527185341e-06,
|
2026 |
-
"loss": 2.0779,
|
2027 |
-
"step": 1660
|
2028 |
-
},
|
2029 |
-
{
|
2030 |
-
"epoch": 0.47,
|
2031 |
-
"learning_rate": 4.752177831180535e-06,
|
2032 |
-
"loss": 2.4298,
|
2033 |
-
"step": 1665
|
2034 |
-
},
|
2035 |
-
{
|
2036 |
-
"epoch": 0.47,
|
2037 |
-
"learning_rate": 4.750300390507661e-06,
|
2038 |
-
"loss": 2.0593,
|
2039 |
-
"step": 1670
|
2040 |
-
},
|
2041 |
-
{
|
2042 |
-
"epoch": 0.47,
|
2043 |
-
"learning_rate": 4.748422949834786e-06,
|
2044 |
-
"loss": 2.0854,
|
2045 |
-
"step": 1675
|
2046 |
-
},
|
2047 |
-
{
|
2048 |
-
"epoch": 0.47,
|
2049 |
-
"learning_rate": 4.746545509161911e-06,
|
2050 |
-
"loss": 2.0465,
|
2051 |
-
"step": 1680
|
2052 |
-
},
|
2053 |
-
{
|
2054 |
-
"epoch": 0.47,
|
2055 |
-
"learning_rate": 4.7446680684890365e-06,
|
2056 |
-
"loss": 2.1006,
|
2057 |
-
"step": 1685
|
2058 |
-
},
|
2059 |
-
{
|
2060 |
-
"epoch": 0.47,
|
2061 |
-
"learning_rate": 4.742790627816161e-06,
|
2062 |
-
"loss": 2.1144,
|
2063 |
-
"step": 1690
|
2064 |
-
},
|
2065 |
-
{
|
2066 |
-
"epoch": 0.47,
|
2067 |
-
"learning_rate": 4.740913187143287e-06,
|
2068 |
-
"loss": 2.355,
|
2069 |
-
"step": 1695
|
2070 |
-
},
|
2071 |
-
{
|
2072 |
-
"epoch": 0.47,
|
2073 |
-
"learning_rate": 4.739035746470412e-06,
|
2074 |
-
"loss": 2.0663,
|
2075 |
-
"step": 1700
|
2076 |
-
},
|
2077 |
-
{
|
2078 |
-
"epoch": 0.48,
|
2079 |
-
"learning_rate": 4.737158305797537e-06,
|
2080 |
-
"loss": 1.9655,
|
2081 |
-
"step": 1705
|
2082 |
-
},
|
2083 |
-
{
|
2084 |
-
"epoch": 0.48,
|
2085 |
-
"learning_rate": 4.735280865124663e-06,
|
2086 |
-
"loss": 1.9764,
|
2087 |
-
"step": 1710
|
2088 |
-
},
|
2089 |
-
{
|
2090 |
-
"epoch": 0.48,
|
2091 |
-
"learning_rate": 4.733403424451788e-06,
|
2092 |
-
"loss": 2.1639,
|
2093 |
-
"step": 1715
|
2094 |
-
},
|
2095 |
-
{
|
2096 |
-
"epoch": 0.48,
|
2097 |
-
"learning_rate": 4.731525983778913e-06,
|
2098 |
-
"loss": 1.9601,
|
2099 |
-
"step": 1720
|
2100 |
-
},
|
2101 |
-
{
|
2102 |
-
"epoch": 0.48,
|
2103 |
-
"learning_rate": 4.729648543106038e-06,
|
2104 |
-
"loss": 2.0582,
|
2105 |
-
"step": 1725
|
2106 |
-
},
|
2107 |
-
{
|
2108 |
-
"epoch": 0.48,
|
2109 |
-
"learning_rate": 4.727771102433164e-06,
|
2110 |
-
"loss": 1.8999,
|
2111 |
-
"step": 1730
|
2112 |
-
},
|
2113 |
-
{
|
2114 |
-
"epoch": 0.48,
|
2115 |
-
"learning_rate": 4.725893661760289e-06,
|
2116 |
-
"loss": 2.1299,
|
2117 |
-
"step": 1735
|
2118 |
-
},
|
2119 |
-
{
|
2120 |
-
"epoch": 0.49,
|
2121 |
-
"learning_rate": 4.724016221087414e-06,
|
2122 |
-
"loss": 1.9507,
|
2123 |
-
"step": 1740
|
2124 |
-
},
|
2125 |
-
{
|
2126 |
-
"epoch": 0.49,
|
2127 |
-
"learning_rate": 4.722138780414539e-06,
|
2128 |
-
"loss": 1.9841,
|
2129 |
-
"step": 1745
|
2130 |
-
},
|
2131 |
-
{
|
2132 |
-
"epoch": 0.49,
|
2133 |
-
"learning_rate": 4.7202613397416646e-06,
|
2134 |
-
"loss": 2.2819,
|
2135 |
-
"step": 1750
|
2136 |
-
},
|
2137 |
-
{
|
2138 |
-
"epoch": 0.49,
|
2139 |
-
"learning_rate": 4.71838389906879e-06,
|
2140 |
-
"loss": 2.0013,
|
2141 |
-
"step": 1755
|
2142 |
-
},
|
2143 |
-
{
|
2144 |
-
"epoch": 0.49,
|
2145 |
-
"learning_rate": 4.716506458395915e-06,
|
2146 |
-
"loss": 1.8868,
|
2147 |
-
"step": 1760
|
2148 |
-
},
|
2149 |
-
{
|
2150 |
-
"epoch": 0.49,
|
2151 |
-
"learning_rate": 4.71462901772304e-06,
|
2152 |
-
"loss": 1.8533,
|
2153 |
-
"step": 1765
|
2154 |
-
},
|
2155 |
-
{
|
2156 |
-
"epoch": 0.49,
|
2157 |
-
"learning_rate": 4.7127515770501655e-06,
|
2158 |
-
"loss": 2.3557,
|
2159 |
-
"step": 1770
|
2160 |
-
},
|
2161 |
-
{
|
2162 |
-
"epoch": 0.5,
|
2163 |
-
"learning_rate": 4.710874136377291e-06,
|
2164 |
-
"loss": 1.7171,
|
2165 |
-
"step": 1775
|
2166 |
-
},
|
2167 |
-
{
|
2168 |
-
"epoch": 0.5,
|
2169 |
-
"learning_rate": 4.708996695704416e-06,
|
2170 |
-
"loss": 1.88,
|
2171 |
-
"step": 1780
|
2172 |
-
},
|
2173 |
-
{
|
2174 |
-
"epoch": 0.5,
|
2175 |
-
"learning_rate": 4.707119255031541e-06,
|
2176 |
-
"loss": 2.1387,
|
2177 |
-
"step": 1785
|
2178 |
-
},
|
2179 |
-
{
|
2180 |
-
"epoch": 0.5,
|
2181 |
-
"learning_rate": 4.7052418143586665e-06,
|
2182 |
-
"loss": 1.8137,
|
2183 |
-
"step": 1790
|
2184 |
-
},
|
2185 |
-
{
|
2186 |
-
"epoch": 0.5,
|
2187 |
-
"learning_rate": 4.703364373685792e-06,
|
2188 |
-
"loss": 1.8639,
|
2189 |
-
"step": 1795
|
2190 |
-
},
|
2191 |
-
{
|
2192 |
-
"epoch": 0.5,
|
2193 |
-
"learning_rate": 4.701486933012917e-06,
|
2194 |
-
"loss": 2.2175,
|
2195 |
-
"step": 1800
|
2196 |
-
},
|
2197 |
-
{
|
2198 |
-
"epoch": 0.5,
|
2199 |
-
"learning_rate": 4.699609492340042e-06,
|
2200 |
-
"loss": 1.7335,
|
2201 |
-
"step": 1805
|
2202 |
-
},
|
2203 |
-
{
|
2204 |
-
"epoch": 0.51,
|
2205 |
-
"learning_rate": 4.6977320516671675e-06,
|
2206 |
-
"loss": 2.1544,
|
2207 |
-
"step": 1810
|
2208 |
-
},
|
2209 |
-
{
|
2210 |
-
"epoch": 0.51,
|
2211 |
-
"learning_rate": 4.6958546109942935e-06,
|
2212 |
-
"loss": 2.1503,
|
2213 |
-
"step": 1815
|
2214 |
-
},
|
2215 |
-
{
|
2216 |
-
"epoch": 0.51,
|
2217 |
-
"learning_rate": 4.693977170321418e-06,
|
2218 |
-
"loss": 2.3852,
|
2219 |
-
"step": 1820
|
2220 |
-
},
|
2221 |
-
{
|
2222 |
-
"epoch": 0.51,
|
2223 |
-
"learning_rate": 4.692099729648544e-06,
|
2224 |
-
"loss": 1.9649,
|
2225 |
-
"step": 1825
|
2226 |
-
},
|
2227 |
-
{
|
2228 |
-
"epoch": 0.51,
|
2229 |
-
"learning_rate": 4.690222288975668e-06,
|
2230 |
-
"loss": 1.9533,
|
2231 |
-
"step": 1830
|
2232 |
-
},
|
2233 |
-
{
|
2234 |
-
"epoch": 0.51,
|
2235 |
-
"learning_rate": 4.688344848302794e-06,
|
2236 |
-
"loss": 1.7411,
|
2237 |
-
"step": 1835
|
2238 |
-
},
|
2239 |
-
{
|
2240 |
-
"epoch": 0.51,
|
2241 |
-
"learning_rate": 4.68646740762992e-06,
|
2242 |
-
"loss": 1.8008,
|
2243 |
-
"step": 1840
|
2244 |
-
},
|
2245 |
-
{
|
2246 |
-
"epoch": 0.52,
|
2247 |
-
"learning_rate": 4.684589966957044e-06,
|
2248 |
-
"loss": 2.0014,
|
2249 |
-
"step": 1845
|
2250 |
-
},
|
2251 |
-
{
|
2252 |
-
"epoch": 0.52,
|
2253 |
-
"learning_rate": 4.68271252628417e-06,
|
2254 |
-
"loss": 1.7596,
|
2255 |
-
"step": 1850
|
2256 |
-
},
|
2257 |
-
{
|
2258 |
-
"epoch": 0.52,
|
2259 |
-
"learning_rate": 4.680835085611295e-06,
|
2260 |
-
"loss": 1.6665,
|
2261 |
-
"step": 1855
|
2262 |
-
},
|
2263 |
-
{
|
2264 |
-
"epoch": 0.52,
|
2265 |
-
"learning_rate": 4.678957644938421e-06,
|
2266 |
-
"loss": 1.6203,
|
2267 |
-
"step": 1860
|
2268 |
-
},
|
2269 |
-
{
|
2270 |
-
"epoch": 0.52,
|
2271 |
-
"learning_rate": 4.677080204265546e-06,
|
2272 |
-
"loss": 2.5619,
|
2273 |
-
"step": 1865
|
2274 |
-
},
|
2275 |
-
{
|
2276 |
-
"epoch": 0.52,
|
2277 |
-
"learning_rate": 4.675202763592671e-06,
|
2278 |
-
"loss": 1.8384,
|
2279 |
-
"step": 1870
|
2280 |
-
},
|
2281 |
-
{
|
2282 |
-
"epoch": 0.52,
|
2283 |
-
"learning_rate": 4.673325322919796e-06,
|
2284 |
-
"loss": 1.7171,
|
2285 |
-
"step": 1875
|
2286 |
-
},
|
2287 |
-
{
|
2288 |
-
"epoch": 0.53,
|
2289 |
-
"learning_rate": 4.671447882246921e-06,
|
2290 |
-
"loss": 1.594,
|
2291 |
-
"step": 1880
|
2292 |
-
},
|
2293 |
-
{
|
2294 |
-
"epoch": 0.53,
|
2295 |
-
"learning_rate": 4.669570441574047e-06,
|
2296 |
-
"loss": 1.5099,
|
2297 |
-
"step": 1885
|
2298 |
-
},
|
2299 |
-
{
|
2300 |
-
"epoch": 0.53,
|
2301 |
-
"learning_rate": 4.667693000901172e-06,
|
2302 |
-
"loss": 1.5688,
|
2303 |
-
"step": 1890
|
2304 |
-
},
|
2305 |
-
{
|
2306 |
-
"epoch": 0.53,
|
2307 |
-
"learning_rate": 4.665815560228297e-06,
|
2308 |
-
"loss": 1.9561,
|
2309 |
-
"step": 1895
|
2310 |
-
},
|
2311 |
-
{
|
2312 |
-
"epoch": 0.53,
|
2313 |
-
"learning_rate": 4.663938119555423e-06,
|
2314 |
-
"loss": 1.7206,
|
2315 |
-
"step": 1900
|
2316 |
-
},
|
2317 |
-
{
|
2318 |
-
"epoch": 0.53,
|
2319 |
-
"learning_rate": 4.662060678882548e-06,
|
2320 |
-
"loss": 1.6661,
|
2321 |
-
"step": 1905
|
2322 |
-
},
|
2323 |
-
{
|
2324 |
-
"epoch": 0.53,
|
2325 |
-
"learning_rate": 4.660183238209673e-06,
|
2326 |
-
"loss": 1.5018,
|
2327 |
-
"step": 1910
|
2328 |
-
},
|
2329 |
-
{
|
2330 |
-
"epoch": 0.54,
|
2331 |
-
"learning_rate": 4.658305797536798e-06,
|
2332 |
-
"loss": 1.7274,
|
2333 |
-
"step": 1915
|
2334 |
-
},
|
2335 |
-
{
|
2336 |
-
"epoch": 0.54,
|
2337 |
-
"learning_rate": 4.656428356863924e-06,
|
2338 |
-
"loss": 1.5007,
|
2339 |
-
"step": 1920
|
2340 |
-
},
|
2341 |
-
{
|
2342 |
-
"epoch": 0.54,
|
2343 |
-
"learning_rate": 4.654550916191049e-06,
|
2344 |
-
"loss": 1.4662,
|
2345 |
-
"step": 1925
|
2346 |
-
},
|
2347 |
-
{
|
2348 |
-
"epoch": 0.54,
|
2349 |
-
"learning_rate": 4.652673475518174e-06,
|
2350 |
-
"loss": 1.4984,
|
2351 |
-
"step": 1930
|
2352 |
-
},
|
2353 |
-
{
|
2354 |
-
"epoch": 0.54,
|
2355 |
-
"learning_rate": 4.650796034845299e-06,
|
2356 |
-
"loss": 1.3899,
|
2357 |
-
"step": 1935
|
2358 |
-
},
|
2359 |
-
{
|
2360 |
-
"epoch": 0.54,
|
2361 |
-
"learning_rate": 4.6489185941724245e-06,
|
2362 |
-
"loss": 1.4297,
|
2363 |
-
"step": 1940
|
2364 |
-
},
|
2365 |
-
{
|
2366 |
-
"epoch": 0.54,
|
2367 |
-
"learning_rate": 4.64704115349955e-06,
|
2368 |
-
"loss": 1.7618,
|
2369 |
-
"step": 1945
|
2370 |
-
},
|
2371 |
-
{
|
2372 |
-
"epoch": 0.54,
|
2373 |
-
"learning_rate": 4.645163712826675e-06,
|
2374 |
-
"loss": 1.6308,
|
2375 |
-
"step": 1950
|
2376 |
-
},
|
2377 |
-
{
|
2378 |
-
"epoch": 0.55,
|
2379 |
-
"learning_rate": 4.6432862721538e-06,
|
2380 |
-
"loss": 1.6705,
|
2381 |
-
"step": 1955
|
2382 |
-
},
|
2383 |
-
{
|
2384 |
-
"epoch": 0.55,
|
2385 |
-
"learning_rate": 4.6414088314809255e-06,
|
2386 |
-
"loss": 1.6243,
|
2387 |
-
"step": 1960
|
2388 |
-
},
|
2389 |
-
{
|
2390 |
-
"epoch": 0.55,
|
2391 |
-
"learning_rate": 4.639531390808051e-06,
|
2392 |
-
"loss": 1.331,
|
2393 |
-
"step": 1965
|
2394 |
-
},
|
2395 |
-
{
|
2396 |
-
"epoch": 0.55,
|
2397 |
-
"learning_rate": 4.637653950135176e-06,
|
2398 |
-
"loss": 1.913,
|
2399 |
-
"step": 1970
|
2400 |
-
},
|
2401 |
-
{
|
2402 |
-
"epoch": 0.55,
|
2403 |
-
"learning_rate": 4.635776509462301e-06,
|
2404 |
-
"loss": 1.6333,
|
2405 |
-
"step": 1975
|
2406 |
-
},
|
2407 |
-
{
|
2408 |
-
"epoch": 0.55,
|
2409 |
-
"learning_rate": 4.6338990687894265e-06,
|
2410 |
-
"loss": 1.4198,
|
2411 |
-
"step": 1980
|
2412 |
-
},
|
2413 |
-
{
|
2414 |
-
"epoch": 0.55,
|
2415 |
-
"learning_rate": 4.632021628116552e-06,
|
2416 |
-
"loss": 2.0835,
|
2417 |
-
"step": 1985
|
2418 |
-
},
|
2419 |
-
{
|
2420 |
-
"epoch": 0.56,
|
2421 |
-
"learning_rate": 4.630144187443677e-06,
|
2422 |
-
"loss": 1.3693,
|
2423 |
-
"step": 1990
|
2424 |
-
},
|
2425 |
-
{
|
2426 |
-
"epoch": 0.56,
|
2427 |
-
"learning_rate": 4.628266746770802e-06,
|
2428 |
-
"loss": 1.5302,
|
2429 |
-
"step": 1995
|
2430 |
-
},
|
2431 |
-
{
|
2432 |
-
"epoch": 0.56,
|
2433 |
-
"learning_rate": 4.6263893060979274e-06,
|
2434 |
-
"loss": 1.8702,
|
2435 |
-
"step": 2000
|
2436 |
-
},
|
2437 |
-
{
|
2438 |
-
"epoch": 0.56,
|
2439 |
-
"eval_loss": 1.0430355072021484,
|
2440 |
-
"eval_runtime": 152.8141,
|
2441 |
-
"eval_samples_per_second": 7.813,
|
2442 |
-
"eval_steps_per_second": 0.982,
|
2443 |
-
"eval_wer": 1.0,
|
2444 |
-
"step": 2000
|
2445 |
-
}
|
2446 |
-
],
|
2447 |
-
"max_steps": 14316,
|
2448 |
-
"num_train_epochs": 4,
|
2449 |
-
"total_flos": 2.7924067835086464e+17,
|
2450 |
-
"trial_name": null,
|
2451 |
-
"trial_params": null
|
2452 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-2000/training_args.bin
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:6f05c6a97d90fd87e90a83547a454ff788a066d3d434f2903b0a9058eae5eb5d
|
3 |
-
size 2991
|
|
|
|
|
|
|
|