Hugging Face's logo

Vyvo
/
echo-ft

Model card Files
xet
Community
kadirnar commited on
Commit
5eebf2d
·
verified ·
1 Parent(s): 13fd25e

Upload checkpoints

Browse files
Files changed (19) hide show
  1. echodit_finetuned/checkpoint-400/flow_model/config.yaml +73 -0
  2. echodit_finetuned/checkpoint-400/flow_model/ema_model.bin +3 -0
  3. echodit_finetuned/checkpoint-400/flow_model/pytorch_model.bin +3 -0
  4. echodit_finetuned/checkpoint-400/flow_model/pytorch_model_ema.bin +3 -0
  5. echodit_finetuned/checkpoint-400/optimizer.pt +3 -0
  6. echodit_finetuned/checkpoint-400/rng_state_0.pth +3 -0
  7. echodit_finetuned/checkpoint-400/rng_state_1.pth +3 -0
  8. echodit_finetuned/checkpoint-400/rng_state_2.pth +3 -0
  9. echodit_finetuned/checkpoint-400/rng_state_3.pth +3 -0
  10. echodit_finetuned/checkpoint-400/rng_state_4.pth +3 -0
  11. echodit_finetuned/checkpoint-400/rng_state_5.pth +3 -0
  12. echodit_finetuned/checkpoint-400/rng_state_6.pth +3 -0
  13. echodit_finetuned/checkpoint-400/rng_state_7.pth +3 -0
  14. echodit_finetuned/checkpoint-400/scheduler.pt +3 -0
  15. echodit_finetuned/checkpoint-400/trainer_state.json +314 -0
  16. echodit_finetuned/final/flow_model/config.yaml +73 -0
  17. echodit_finetuned/final/flow_model/ema_model.bin +3 -0
  18. echodit_finetuned/final/flow_model/pytorch_model.bin +3 -0
  19. echodit_finetuned/final/flow_model/pytorch_model_ema.bin +3 -0
echodit_finetuned/checkpoint-400/flow_model/config.yaml ADDED
@@ -0,0 +1,73 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ TTS_dataset_local: ./finetune_data
2
+ adaln_rank: 128
3
+ batch_size: 16
4
+ cfg_dropout: 0.1
5
+ dacvae_latent_dim: 128
6
+ dacvae_model: facebook/dacvae-watermarked
7
+ dataloader_drop_last: true
8
+ dataloader_num_workers: 4
9
+ dataloader_pin_memory: true
10
+ do_validation: false
11
+ ema_decay: 0.9999
12
+ ema_update_every: 10
13
+ end_of_ai: 100285
14
+ end_of_human: 100283
15
+ end_of_speech: 100281
16
+ end_of_text: 100279
17
+ epochs: 100
18
+ eval_steps: 500
19
+ finetune_mode: full
20
+ flow_sigma_min: 0.0001
21
+ flow_velocity_weighted: false
22
+ freeze_first_n_layers: 0
23
+ freeze_speaker_encoder: true
24
+ freeze_text_encoder: false
25
+ gradient_accumulation_steps: 4
26
+ gradient_checkpointing: true
27
+ intermediate_size: 4096
28
+ learning_rate: 1.0e-05
29
+ logging_steps: 10
30
+ lr_min_ratio: 0.1
31
+ lr_scheduler_type: cosine
32
+ max_grad_norm: 0.5
33
+ mixed_precision: bf16
34
+ model_size: 1024
35
+ norm_eps: 1.0e-06
36
+ num_heads: 16
37
+ num_layers: 24
38
+ number_processes: 8
39
+ pad_token: 100286
40
+ pretrained_checkpoint: /scratch/huggingface/emilia-yodas-dacvae_tokenized/checkpoints/echodit_1b/final/flow_model/pytorch_model.bin
41
+ project_name: vyvo-echodit-finetune
42
+ resume_from_checkpoint: null
43
+ run_name: echodit-finetune-andrew
44
+ save_folder: checkpoints/echodit_finetuned
45
+ save_steps: 1000
46
+ speaker_intermediate_size: 2048
47
+ speaker_model_size: 512
48
+ speaker_num_heads: 8
49
+ speaker_num_layers: 4
50
+ speaker_patch_size: 4
51
+ start_of_ai: 100284
52
+ start_of_human: 100282
53
+ start_of_speech: 100280
54
+ start_of_text: 100278
55
+ text_intermediate_size: 3072
56
+ text_model_size: 768
57
+ text_num_heads: 12
58
+ text_num_layers: 6
59
+ text_vocab_size: 100287
60
+ timestep_embed_size: 256
61
+ tokeniser_length: 100277
62
+ unfreeze_last_n_layers: 24
63
+ use_ema: true
64
+ use_flash_attention: true
65
+ use_fsdp: false
66
+ validation_cfg_scale: 2.5
67
+ validation_ode_steps: 50
68
+ validation_samples: 4
69
+ validation_solver: heun
70
+ validation_steps: 500
71
+ warmup_ratio: 0.05
72
+ warmup_steps: 500
73
+ weight_decay: 0.01
echodit_finetuned/checkpoint-400/flow_model/ema_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0aa9e64c1635489ed6e0ea1b6af6a5290cd7a8066c28e7c0103053d488d9c087
3
+ size 2855777547
echodit_finetuned/checkpoint-400/flow_model/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aac268b20e8ab41765c59138478a3b472e5047a91a393552e37ec943f365bc7e
3
+ size 2928174600
echodit_finetuned/checkpoint-400/flow_model/pytorch_model_ema.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8744bf8629d842c1e8b0ddc7e92877a5289a9fa193fb15defce411d75c462403
3
+ size 2855771883
echodit_finetuned/checkpoint-400/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9f3eb5db843f26c995c1a9c23ae247d0efb6011b0d4168d5dbef41c0211bc5fc
3
+ size 5365560092
echodit_finetuned/checkpoint-400/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ccc4361efca7928a41093bafd03808a06414df27e867b11e3c143637a21e3397
3
+ size 16389
echodit_finetuned/checkpoint-400/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5cb3b39fb9b516e701b3cadd7af82628d96076624a4124819ff69ed4542c07e7
3
+ size 16389
echodit_finetuned/checkpoint-400/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:79be7b437a5dbff8167ead07fce81738c149576339b6257a15504c24f12fa1b0
3
+ size 16389
echodit_finetuned/checkpoint-400/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2fc5801bef714d4f8330418056d8138f05b9434b09dbfacce531ee69bc1616de
3
+ size 16389
echodit_finetuned/checkpoint-400/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eb709ce9638db60d967e2c7f5eb34b5f0f867a4aaeabe34a2c4f79fe7c7eae3a
3
+ size 16389
echodit_finetuned/checkpoint-400/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ab41712355ce3c8adb44e543fe8cbd921ad289585b4961ecbb4ca96ac7cb0b72
3
+ size 16389
echodit_finetuned/checkpoint-400/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3b4a2d54e47899c49948f737b6a1d2c46882cb02b43cd494bd29c58616884d44
3
+ size 16389
echodit_finetuned/checkpoint-400/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:319f7332a6384b8d779ea942b11f3ac530c01be9acd0181046e11fd67cdd5bd0
3
+ size 16389
echodit_finetuned/checkpoint-400/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a39f2b6581e41979c4ab06dc53a5045fe90458186744d73a43c93000caef346b
3
+ size 1465
echodit_finetuned/checkpoint-400/trainer_state.json ADDED
@@ -0,0 +1,314 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 100.0,
6
+ "eval_steps": 500,
7
+ "global_step": 400,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 2.533333333333333,
14
+ "grad_norm": 0.5185111165046692,
15
+ "learning_rate": 1.8e-07,
16
+ "loss": 0.5955,
17
+ "step": 10
18
+ },
19
+ {
20
+ "epoch": 5.0,
21
+ "grad_norm": 0.5055148005485535,
22
+ "learning_rate": 3.8e-07,
23
+ "loss": 0.5933,
24
+ "step": 20
25
+ },
26
+ {
27
+ "epoch": 7.533333333333333,
28
+ "grad_norm": 0.45245474576950073,
29
+ "learning_rate": 5.800000000000001e-07,
30
+ "loss": 0.5906,
31
+ "step": 30
32
+ },
33
+ {
34
+ "epoch": 10.0,
35
+ "grad_norm": 0.4057580530643463,
36
+ "learning_rate": 7.8e-07,
37
+ "loss": 0.5873,
38
+ "step": 40
39
+ },
40
+ {
41
+ "epoch": 12.533333333333333,
42
+ "grad_norm": 0.34415045380592346,
43
+ "learning_rate": 9.800000000000001e-07,
44
+ "loss": 0.5817,
45
+ "step": 50
46
+ },
47
+ {
48
+ "epoch": 15.0,
49
+ "grad_norm": 0.2918901741504669,
50
+ "learning_rate": 1.1800000000000001e-06,
51
+ "loss": 0.5754,
52
+ "step": 60
53
+ },
54
+ {
55
+ "epoch": 17.533333333333335,
56
+ "grad_norm": 0.22454246878623962,
57
+ "learning_rate": 1.3800000000000001e-06,
58
+ "loss": 0.5739,
59
+ "step": 70
60
+ },
61
+ {
62
+ "epoch": 20.0,
63
+ "grad_norm": 0.2028179168701172,
64
+ "learning_rate": 1.5800000000000001e-06,
65
+ "loss": 0.5675,
66
+ "step": 80
67
+ },
68
+ {
69
+ "epoch": 22.533333333333335,
70
+ "grad_norm": 0.15407922863960266,
71
+ "learning_rate": 1.7800000000000001e-06,
72
+ "loss": 0.5644,
73
+ "step": 90
74
+ },
75
+ {
76
+ "epoch": 25.0,
77
+ "grad_norm": 0.15080758929252625,
78
+ "learning_rate": 1.98e-06,
79
+ "loss": 0.5607,
80
+ "step": 100
81
+ },
82
+ {
83
+ "epoch": 27.533333333333335,
84
+ "grad_norm": 0.14132018387317657,
85
+ "learning_rate": 2.1800000000000003e-06,
86
+ "loss": 0.5584,
87
+ "step": 110
88
+ },
89
+ {
90
+ "epoch": 30.0,
91
+ "grad_norm": 0.1372448056936264,
92
+ "learning_rate": 2.38e-06,
93
+ "loss": 0.5542,
94
+ "step": 120
95
+ },
96
+ {
97
+ "epoch": 32.53333333333333,
98
+ "grad_norm": 0.1235460564494133,
99
+ "learning_rate": 2.5800000000000003e-06,
100
+ "loss": 0.5527,
101
+ "step": 130
102
+ },
103
+ {
104
+ "epoch": 35.0,
105
+ "grad_norm": 0.12351451069116592,
106
+ "learning_rate": 2.7800000000000005e-06,
107
+ "loss": 0.5514,
108
+ "step": 140
109
+ },
110
+ {
111
+ "epoch": 37.53333333333333,
112
+ "grad_norm": 0.16762301325798035,
113
+ "learning_rate": 2.9800000000000003e-06,
114
+ "loss": 0.5515,
115
+ "step": 150
116
+ },
117
+ {
118
+ "epoch": 40.0,
119
+ "grad_norm": 0.121138796210289,
120
+ "learning_rate": 3.1800000000000005e-06,
121
+ "loss": 0.5487,
122
+ "step": 160
123
+ },
124
+ {
125
+ "epoch": 42.53333333333333,
126
+ "grad_norm": 0.1075759157538414,
127
+ "learning_rate": 3.3800000000000007e-06,
128
+ "loss": 0.5434,
129
+ "step": 170
130
+ },
131
+ {
132
+ "epoch": 45.0,
133
+ "grad_norm": 0.11517618596553802,
134
+ "learning_rate": 3.58e-06,
135
+ "loss": 0.5455,
136
+ "step": 180
137
+ },
138
+ {
139
+ "epoch": 47.53333333333333,
140
+ "grad_norm": 0.10481545329093933,
141
+ "learning_rate": 3.7800000000000002e-06,
142
+ "loss": 0.5446,
143
+ "step": 190
144
+ },
145
+ {
146
+ "epoch": 50.0,
147
+ "grad_norm": 0.13033606112003326,
148
+ "learning_rate": 3.980000000000001e-06,
149
+ "loss": 0.5426,
150
+ "step": 200
151
+ },
152
+ {
153
+ "epoch": 52.53333333333333,
154
+ "grad_norm": 0.11076183617115021,
155
+ "learning_rate": 4.18e-06,
156
+ "loss": 0.5425,
157
+ "step": 210
158
+ },
159
+ {
160
+ "epoch": 55.0,
161
+ "grad_norm": 0.13085302710533142,
162
+ "learning_rate": 4.38e-06,
163
+ "loss": 0.5407,
164
+ "step": 220
165
+ },
166
+ {
167
+ "epoch": 57.53333333333333,
168
+ "grad_norm": 0.1101340726017952,
169
+ "learning_rate": 4.58e-06,
170
+ "loss": 0.5393,
171
+ "step": 230
172
+ },
173
+ {
174
+ "epoch": 60.0,
175
+ "grad_norm": 0.11731196194887161,
176
+ "learning_rate": 4.78e-06,
177
+ "loss": 0.5373,
178
+ "step": 240
179
+ },
180
+ {
181
+ "epoch": 62.53333333333333,
182
+ "grad_norm": 0.11329876631498337,
183
+ "learning_rate": 4.980000000000001e-06,
184
+ "loss": 0.5353,
185
+ "step": 250
186
+ },
187
+ {
188
+ "epoch": 65.0,
189
+ "grad_norm": 0.12666642665863037,
190
+ "learning_rate": 5.18e-06,
191
+ "loss": 0.5365,
192
+ "step": 260
193
+ },
194
+ {
195
+ "epoch": 67.53333333333333,
196
+ "grad_norm": 0.12180330604314804,
197
+ "learning_rate": 5.380000000000001e-06,
198
+ "loss": 0.5353,
199
+ "step": 270
200
+ },
201
+ {
202
+ "epoch": 70.0,
203
+ "grad_norm": 0.11484792828559875,
204
+ "learning_rate": 5.580000000000001e-06,
205
+ "loss": 0.5309,
206
+ "step": 280
207
+ },
208
+ {
209
+ "epoch": 72.53333333333333,
210
+ "grad_norm": 0.10521823912858963,
211
+ "learning_rate": 5.78e-06,
212
+ "loss": 0.5336,
213
+ "step": 290
214
+ },
215
+ {
216
+ "epoch": 75.0,
217
+ "grad_norm": 0.12647727131843567,
218
+ "learning_rate": 5.98e-06,
219
+ "loss": 0.533,
220
+ "step": 300
221
+ },
222
+ {
223
+ "epoch": 77.53333333333333,
224
+ "grad_norm": 0.11960894614458084,
225
+ "learning_rate": 6.18e-06,
226
+ "loss": 0.5305,
227
+ "step": 310
228
+ },
229
+ {
230
+ "epoch": 80.0,
231
+ "grad_norm": 0.14461685717105865,
232
+ "learning_rate": 6.380000000000001e-06,
233
+ "loss": 0.5297,
234
+ "step": 320
235
+ },
236
+ {
237
+ "epoch": 82.53333333333333,
238
+ "grad_norm": 0.11162838339805603,
239
+ "learning_rate": 6.5800000000000005e-06,
240
+ "loss": 0.5309,
241
+ "step": 330
242
+ },
243
+ {
244
+ "epoch": 85.0,
245
+ "grad_norm": 0.12640050053596497,
246
+ "learning_rate": 6.780000000000001e-06,
247
+ "loss": 0.5264,
248
+ "step": 340
249
+ },
250
+ {
251
+ "epoch": 87.53333333333333,
252
+ "grad_norm": 0.11897486448287964,
253
+ "learning_rate": 6.98e-06,
254
+ "loss": 0.5267,
255
+ "step": 350
256
+ },
257
+ {
258
+ "epoch": 90.0,
259
+ "grad_norm": 0.11157895624637604,
260
+ "learning_rate": 7.180000000000001e-06,
261
+ "loss": 0.5273,
262
+ "step": 360
263
+ },
264
+ {
265
+ "epoch": 92.53333333333333,
266
+ "grad_norm": 0.11413957178592682,
267
+ "learning_rate": 7.3800000000000005e-06,
268
+ "loss": 0.527,
269
+ "step": 370
270
+ },
271
+ {
272
+ "epoch": 95.0,
273
+ "grad_norm": 0.12203444540500641,
274
+ "learning_rate": 7.58e-06,
275
+ "loss": 0.5258,
276
+ "step": 380
277
+ },
278
+ {
279
+ "epoch": 97.53333333333333,
280
+ "grad_norm": 0.10855170339345932,
281
+ "learning_rate": 7.78e-06,
282
+ "loss": 0.5255,
283
+ "step": 390
284
+ },
285
+ {
286
+ "epoch": 100.0,
287
+ "grad_norm": 0.13235536217689514,
288
+ "learning_rate": 7.980000000000002e-06,
289
+ "loss": 0.5228,
290
+ "step": 400
291
+ }
292
+ ],
293
+ "logging_steps": 10,
294
+ "max_steps": 400,
295
+ "num_input_tokens_seen": 0,
296
+ "num_train_epochs": 100,
297
+ "save_steps": 1000,
298
+ "stateful_callbacks": {
299
+ "TrainerControl": {
300
+ "args": {
301
+ "should_epoch_stop": false,
302
+ "should_evaluate": false,
303
+ "should_log": false,
304
+ "should_save": true,
305
+ "should_training_stop": true
306
+ },
307
+ "attributes": {}
308
+ }
309
+ },
310
+ "total_flos": 0.0,
311
+ "train_batch_size": 16,
312
+ "trial_name": null,
313
+ "trial_params": null
314
+ }
echodit_finetuned/final/flow_model/config.yaml ADDED
@@ -0,0 +1,73 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ TTS_dataset_local: ./finetune_data
2
+ adaln_rank: 128
3
+ batch_size: 16
4
+ cfg_dropout: 0.1
5
+ dacvae_latent_dim: 128
6
+ dacvae_model: facebook/dacvae-watermarked
7
+ dataloader_drop_last: true
8
+ dataloader_num_workers: 4
9
+ dataloader_pin_memory: true
10
+ do_validation: false
11
+ ema_decay: 0.9999
12
+ ema_update_every: 10
13
+ end_of_ai: 100285
14
+ end_of_human: 100283
15
+ end_of_speech: 100281
16
+ end_of_text: 100279
17
+ epochs: 100
18
+ eval_steps: 500
19
+ finetune_mode: full
20
+ flow_sigma_min: 0.0001
21
+ flow_velocity_weighted: false
22
+ freeze_first_n_layers: 0
23
+ freeze_speaker_encoder: true
24
+ freeze_text_encoder: false
25
+ gradient_accumulation_steps: 4
26
+ gradient_checkpointing: true
27
+ intermediate_size: 4096
28
+ learning_rate: 1.0e-05
29
+ logging_steps: 10
30
+ lr_min_ratio: 0.1
31
+ lr_scheduler_type: cosine
32
+ max_grad_norm: 0.5
33
+ mixed_precision: bf16
34
+ model_size: 1024
35
+ norm_eps: 1.0e-06
36
+ num_heads: 16
37
+ num_layers: 24
38
+ number_processes: 8
39
+ pad_token: 100286
40
+ pretrained_checkpoint: /scratch/huggingface/emilia-yodas-dacvae_tokenized/checkpoints/echodit_1b/final/flow_model/pytorch_model.bin
41
+ project_name: vyvo-echodit-finetune
42
+ resume_from_checkpoint: null
43
+ run_name: echodit-finetune-andrew
44
+ save_folder: checkpoints/echodit_finetuned
45
+ save_steps: 1000
46
+ speaker_intermediate_size: 2048
47
+ speaker_model_size: 512
48
+ speaker_num_heads: 8
49
+ speaker_num_layers: 4
50
+ speaker_patch_size: 4
51
+ start_of_ai: 100284
52
+ start_of_human: 100282
53
+ start_of_speech: 100280
54
+ start_of_text: 100278
55
+ text_intermediate_size: 3072
56
+ text_model_size: 768
57
+ text_num_heads: 12
58
+ text_num_layers: 6
59
+ text_vocab_size: 100287
60
+ timestep_embed_size: 256
61
+ tokeniser_length: 100277
62
+ unfreeze_last_n_layers: 24
63
+ use_ema: true
64
+ use_flash_attention: true
65
+ use_fsdp: false
66
+ validation_cfg_scale: 2.5
67
+ validation_ode_steps: 50
68
+ validation_samples: 4
69
+ validation_solver: heun
70
+ validation_steps: 500
71
+ warmup_ratio: 0.05
72
+ warmup_steps: 500
73
+ weight_decay: 0.01
echodit_finetuned/final/flow_model/ema_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0aa9e64c1635489ed6e0ea1b6af6a5290cd7a8066c28e7c0103053d488d9c087
3
+ size 2855777547
echodit_finetuned/final/flow_model/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aac268b20e8ab41765c59138478a3b472e5047a91a393552e37ec943f365bc7e
3
+ size 2928174600
echodit_finetuned/final/flow_model/pytorch_model_ema.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8744bf8629d842c1e8b0ddc7e92877a5289a9fa193fb15defce411d75c462403
3
+ size 2855771883