Saving train state of step 1000

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "openai/whisper-base",
+  "_name_or_path": "openai/whisper-tiny",
   "activation_dropout": 0.0,
   "activation_function": "gelu",
   "apply_spec_augment": false,
@@ -13,17 +13,17 @@
   ],
   "bos_token_id": 50257,
   "classifier_proj_size": 256,
-  "d_model": 512,
-  "decoder_attention_heads": 8,
-  "decoder_ffn_dim": 2048,
+  "d_model": 384,
+  "decoder_attention_heads": 6,
+  "decoder_ffn_dim": 1536,
   "decoder_layerdrop": 0.0,
-  "decoder_layers": 6,
+  "decoder_layers": 4,
   "decoder_start_token_id": 50258,
   "dropout": 0.0,
-  "encoder_attention_heads": 8,
-  "encoder_ffn_dim": 2048,
+  "encoder_attention_heads": 6,
+  "encoder_ffn_dim": 1536,
   "encoder_layerdrop": 0.0,
-  "encoder_layers": 6,
+  "encoder_layers": 4,
   "eos_token_id": 50257,
   "forced_decoder_ids": [
     [
@@ -52,7 +52,7 @@
   "max_target_positions": 448,
   "median_filter_width": 7,
   "model_type": "whisper",
-  "num_hidden_layers": 6,
+  "num_hidden_layers": 4,
   "num_mel_bins": 80,
   "pad_token_id": 50257,
   "scale_embedding": false,

distil-whisper/events.out.tfevents.1712414653.train02.3753928.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e83f1cd21a4e6043c2334dce2bfa96ecf8cb0d84dcb2d86fcbd7d1775027605f
-size 63446
+oid sha256:b51a92acd49bca43622c5913fb11ace5cd11f669217a6686bea3d17315306783
+size 63793

distil-whisper/events.out.tfevents.1712421884.train02.3768589.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fc0c06b7543dfcbb3a6135880755720796f84a317e6e45d303bc1d8515193bba
+size 12458

dvclive/metrics.json

@@ -1,14 +1,8 @@
 {
     "train": {
-        "time": 2253.3530826568604,
-        "epoch": 5,
+        "time": 356.29300141334534,
+        "epoch": 1,
         "learning_rate": 0.0001
     },
-    "step": 5000,
-    "eval": {
-        "wer": 22.24298186860319,
-        "wer_ortho": 28.8339752069018,
-        "time": 245.54304361343384,
-        "epoch": 4
-    }
+    "step": 1000
 }

dvclive/plots/metrics/train/epoch.tsv

@@ -39,163 +39,3 @@ step	epoch
 950	1
 975	1
 1000	1
-1025	1
-1050	1
-1075	1
-1100	1
-1125	1
-1150	1
-1175	1
-1200	1
-1225	1
-1250	1
-1275	1
-1300	1
-1325	1
-1350	1
-1375	1
-1400	1
-1425	1
-1450	1
-1475	1
-1500	1
-1525	1
-1550	1
-1575	1
-1600	1
-1625	1
-1650	1
-1675	1
-1700	1
-1725	2
-1750	2
-1775	2
-1800	2
-1825	2
-1850	2
-1875	2
-1900	2
-1925	2
-1950	2
-1975	2
-2000	2
-2025	2
-2050	2
-2075	2
-2100	2
-2125	2
-2150	2
-2175	2
-2200	2
-2225	2
-2250	2
-2275	2
-2300	2
-2325	2
-2350	2
-2375	2
-2400	2
-2425	2
-2450	2
-2475	2
-2500	2
-2525	2
-2550	2
-2575	2
-2600	3
-2625	3
-2650	3
-2675	3
-2700	3
-2725	3
-2750	3
-2775	3
-2800	3
-2825	3
-2850	3
-2875	3
-2900	3
-2925	3
-2950	3
-2975	3
-3000	3
-3025	3
-3050	3
-3075	3
-3100	3
-3125	3
-3150	3
-3175	3
-3200	3
-3225	3
-3250	3
-3275	3
-3300	3
-3325	3
-3350	3
-3375	3
-3400	3
-3425	3
-3450	4
-3475	4
-3500	4
-3525	4
-3550	4
-3575	4
-3600	4
-3625	4
-3650	4
-3675	4
-3700	4
-3725	4
-3750	4
-3775	4
-3800	4
-3825	4
-3850	4
-3875	4
-3900	4
-3925	4
-3950	4
-3975	4
-4000	4
-4025	4
-4050	4
-4075	4
-4100	4
-4125	4
-4150	4
-4175	4
-4200	4
-4225	4
-4250	4
-4275	4
-4300	4
-4325	5
-4350	5
-4375	5
-4400	5
-4425	5
-4450	5
-4475	5
-4500	5
-4525	5
-4550	5
-4575	5
-4600	5
-4625	5
-4650	5
-4675	5
-4700	5
-4725	5
-4750	5
-4775	5
-4800	5
-4825	5
-4850	5
-4875	5
-4900	5
-4925	5
-4950	5
-4975	5
-5000	5

dvclive/plots/metrics/train/learning_rate.tsv

@@ -1,6 +1,6 @@
 step	learning_rate
-25	4.4000000000000006e-05
-50	9.4e-05
+25	4e-05
+50	9e-05
 75	0.0001
 100	0.0001
 125	0.0001
@@ -39,163 +39,3 @@ step	learning_rate
 950	0.0001
 975	0.0001
 1000	0.0001
-1025	0.0001
-1050	0.0001
-1075	0.0001
-1100	0.0001
-1125	0.0001
-1150	0.0001
-1175	0.0001
-1200	0.0001
-1225	0.0001
-1250	0.0001
-1275	0.0001
-1300	0.0001
-1325	0.0001
-1350	0.0001
-1375	0.0001
-1400	0.0001
-1425	0.0001
-1450	0.0001
-1475	0.0001
-1500	0.0001
-1525	0.0001
-1550	0.0001
-1575	0.0001
-1600	0.0001
-1625	0.0001
-1650	0.0001
-1675	0.0001
-1700	0.0001
-1725	0.0001
-1750	0.0001
-1775	0.0001
-1800	0.0001
-1825	0.0001
-1850	0.0001
-1875	0.0001
-1900	0.0001
-1925	0.0001
-1950	0.0001
-1975	0.0001
-2000	0.0001
-2025	0.0001
-2050	0.0001
-2075	0.0001
-2100	0.0001
-2125	0.0001
-2150	0.0001
-2175	0.0001
-2200	0.0001
-2225	0.0001
-2250	0.0001
-2275	0.0001
-2300	0.0001
-2325	0.0001
-2350	0.0001
-2375	0.0001
-2400	0.0001
-2425	0.0001
-2450	0.0001
-2475	0.0001
-2500	0.0001
-2525	0.0001
-2550	0.0001
-2575	0.0001
-2600	0.0001
-2625	0.0001
-2650	0.0001
-2675	0.0001
-2700	0.0001
-2725	0.0001
-2750	0.0001
-2775	0.0001
-2800	0.0001
-2825	0.0001
-2850	0.0001
-2875	0.0001
-2900	0.0001
-2925	0.0001
-2950	0.0001
-2975	0.0001
-3000	0.0001
-3025	0.0001
-3050	0.0001
-3075	0.0001
-3100	0.0001
-3125	0.0001
-3150	0.0001
-3175	0.0001
-3200	0.0001
-3225	0.0001
-3250	0.0001
-3275	0.0001
-3300	0.0001
-3325	0.0001
-3350	0.0001
-3375	0.0001
-3400	0.0001
-3425	0.0001
-3450	0.0001
-3475	0.0001
-3500	0.0001
-3525	0.0001
-3550	0.0001
-3575	0.0001
-3600	0.0001
-3625	0.0001
-3650	0.0001
-3675	0.0001
-3700	0.0001
-3725	0.0001
-3750	0.0001
-3775	0.0001
-3800	0.0001
-3825	0.0001
-3850	0.0001
-3875	0.0001
-3900	0.0001
-3925	0.0001
-3950	0.0001
-3975	0.0001
-4000	0.0001
-4025	0.0001
-4050	0.0001
-4075	0.0001
-4100	0.0001
-4125	0.0001
-4150	0.0001
-4175	0.0001
-4200	0.0001
-4225	0.0001
-4250	0.0001
-4275	0.0001
-4300	0.0001
-4325	0.0001
-4350	0.0001
-4375	0.0001
-4400	0.0001
-4425	0.0001
-4450	0.0001
-4475	0.0001
-4500	0.0001
-4525	0.0001
-4550	0.0001
-4575	0.0001
-4600	0.0001
-4625	0.0001
-4650	0.0001
-4675	0.0001
-4700	0.0001
-4725	0.0001
-4750	0.0001
-4775	0.0001
-4800	0.0001
-4825	0.0001
-4850	0.0001
-4875	0.0001
-4900	0.0001
-4925	0.0001
-4950	0.0001
-4975	0.0001
-5000	0.0001

dvclive/plots/metrics/train/time.tsv

@@ -1,201 +1,41 @@
 step	time
-25	11.684435844421387
-50	21.286893844604492
-75	30.893161058425903
-100	40.50417184829712
-125	50.126877784729004
-150	59.74510478973389
-175	69.3599362373352
-200	78.9830493927002
-225	88.60592222213745
-250	98.23294377326965
-275	107.85366106033325
-300	117.4715666770935
-325	127.09286832809448
-350	136.72641706466675
-375	146.37023186683655
-400	156.02068185806274
-425	165.67155933380127
-450	175.32003140449524
-475	184.960928440094
-500	194.60969734191895
-525	204.2479236125946
-550	213.8882131576538
-575	223.52437162399292
-600	233.17730355262756
-625	242.82540273666382
-650	252.46957540512085
-675	262.1174519062042
-700	271.77419233322144
-725	281.43049025535583
-750	291.0802092552185
-775	300.7343370914459
-800	310.38205337524414
-825	320.03323912620544
-850	329.6830699443817
-875	341.0324788093567
-900	350.6875424385071
-925	360.3380663394928
-950	369.98790550231934
-975	379.62979459762573
-1000	389.28764247894287
-1025	477.43934750556946
-1050	487.06905603408813
-1075	496.70161390304565
-1100	506.34014201164246
-1125	515.968909740448
-1150	525.6247124671936
-1175	535.2812929153442
-1200	544.94753074646
-1225	554.6170375347137
-1250	564.2695646286011
-1275	573.9344782829285
-1300	583.5934343338013
-1325	593.2619092464447
-1350	602.9200406074524
-1375	612.5834064483643
-1400	622.2427930831909
-1425	631.9039030075073
-1450	641.5672912597656
-1475	651.2363836765289
-1500	660.9025223255157
-1525	670.5803966522217
-1550	680.2368001937866
-1575	689.8935434818268
-1600	699.561598777771
-1625	709.2326860427856
-1650	718.9015653133392
-1675	728.5651636123657
-1700	738.2388761043549
-1725	749.058812379837
-1750	758.715169429779
-1775	768.3791604042053
-1800	778.0330898761749
-1825	787.7043738365173
-1850	797.3711838722229
-1875	807.033620595932
-1900	816.7156624794006
-1925	826.369535446167
-1950	836.0318987369537
-1975	845.6994693279266
-2000	855.3777701854706
-2025	945.9205918312073
-2050	955.5570874214172
-2075	965.183685541153
-2100	974.8230764865875
-2125	984.4629333019257
-2150	994.1015644073486
-2175	1003.7421622276306
-2200	1013.3943798542023
-2225	1023.0340690612793
-2250	1032.6920757293701
-2275	1042.355932712555
-2300	1052.0133595466614
-2325	1061.6764154434204
-2350	1071.3519291877747
-2375	1081.009160041809
-2400	1090.6628205776215
-2425	1100.3322820663452
-2450	1109.9939272403717
-2475	1119.6580357551575
-2500	1129.3195736408234
-2525	1138.9884753227234
-2550	1148.6439034938812
-2575	1158.2940447330475
-2600	1169.1666514873505
-2625	1178.8291101455688
-2650	1188.4843485355377
-2675	1198.1412012577057
-2700	1207.7901220321655
-2725	1217.4453997612
-2750	1227.106692790985
-2775	1236.758858203888
-2800	1246.4241180419922
-2825	1256.0781526565552
-2850	1265.7430965900421
-2875	1275.3995866775513
-2900	1285.067929983139
-2925	1294.7319355010986
-2950	1304.398491859436
-2975	1314.061797618866
-3000	1323.7229840755463
-3025	1409.9547500610352
-3050	1419.5985000133514
-3075	1429.2499001026154
-3100	1438.9136247634888
-3125	1448.574645280838
-3150	1458.2276797294617
-3175	1467.8867342472076
-3200	1477.555083990097
-3225	1487.21031498909
-3250	1496.8673069477081
-3275	1506.5304148197174
-3300	1516.191492319107
-3325	1525.8438050746918
-3350	1535.5006895065308
-3375	1545.1577517986298
-3400	1554.8170964717865
-3425	1564.486836194992
-3450	1575.3070306777954
-3475	1584.952998638153
-3500	1594.5967934131622
-3525	1604.240556716919
-3550	1613.887773990631
-3575	1623.5305771827698
-3600	1633.1819751262665
-3625	1642.8412942886353
-3650	1652.4885885715485
-3675	1662.1482954025269
-3700	1671.7933938503265
-3725	1681.4414370059967
-3750	1691.099225282669
-3775	1700.7563977241516
-3800	1710.4122366905212
-3825	1720.0654191970825
-3850	1729.7193558216095
-3875	1739.3637006282806
-3900	1749.0059382915497
-3925	1758.6463613510132
-3950	1768.295547246933
-3975	1777.9552805423737
-4000	1787.6112430095673
-4025	1875.8111493587494
-4050	1885.4643814563751
-4075	1895.1054847240448
-4100	1904.7525160312653
-4125	1914.390187740326
-4150	1924.0471398830414
-4175	1933.6979429721832
-4200	1943.349021911621
-4225	1952.9981617927551
-4250	1962.63041639328
-4275	1972.275469303131
-4300	1981.9206290245056
-4325	1992.7996625900269
-4350	2002.4559121131897
-4375	2012.1016597747803
-4400	2021.749123096466
-4425	2031.40646982193
-4450	2041.0575802326202
-4475	2050.7022540569305
-4500	2060.351977586746
-4525	2069.997896671295
-4550	2079.6516761779785
-4575	2089.2996559143066
-4600	2098.954682826996
-4625	2108.6132814884186
-4650	2118.261932849884
-4675	2127.917363882065
-4700	2137.573076248169
-4725	2147.217692375183
-4750	2156.8727049827576
-4775	2166.5152747631073
-4800	2176.1649520397186
-4825	2185.8083612918854
-4850	2195.4502625465393
-4875	2205.109563589096
-4900	2214.761839389801
-4925	2224.407284975052
-4950	2234.056025505066
-4975	2243.701906681061
-5000	2253.3530826568604
+25	11.024884700775146
+50	19.85481309890747
+75	28.683847665786743
+100	37.506720781326294
+125	46.30107140541077
+150	55.08619236946106
+175	63.87367296218872
+200	72.66220188140869
+225	81.45210266113281
+250	90.23731517791748
+275	99.02079439163208
+300	107.81453824043274
+325	116.60759544372559
+350	125.42598938941956
+375	134.24483466148376
+400	143.07194328308105
+425	151.8877546787262
+450	160.70777201652527
+475	169.5274999141693
+500	178.3484025001526
+525	187.16752195358276
+550	196.0018846988678
+575	204.82303142547607
+600	213.64335584640503
+625	222.46671533584595
+650	231.29865288734436
+675	240.13090705871582
+700	248.95567631721497
+725	257.780916929245
+750	266.6054456233978
+775	275.4310476779938
+800	284.2579245567322
+825	293.0841419696808
+850	301.9183979034424
+875	312.1040503978729
+900	320.93589448928833
+925	329.77263498306274
+950	338.6202914714813
+975	347.453581571579
+1000	356.29300141334534

generation_config.json

@@ -1,36 +1,28 @@
 {
   "alignment_heads": [
     [
-      3,
-      1
-    ],
-    [
-      4,
+      2,
       2
     ],
     [
-      4,
-      3
-    ],
-    [
-      4,
-      7
+      3,
+      0
     ],
     [
-      5,
-      1
+      3,
+      2
     ],
     [
-      5,
-      2
+      3,
+      3
     ],
     [
-      5,
+      3,
       4
     ],
     [
-      5,
-      6
+      3,
+      5
     ]
   ],
   "begin_suppress_tokens": [