Final model for experiment Serbian

README.md

@@ -1,7 +1,7 @@
 ---
 library_name: transformers
 tags:
-- Welsh
+- Serbian
 - generated_from_trainer
 model-index:
 - name: childes-segmentation-100k-gpt2_lm-model
@@ -16,53 +16,53 @@ should probably proofread and complete it, then remove this comment. -->
 This model is a fine-tuned version of [](https://huggingface.co/) on an unknown dataset.
 It achieves the following results on the evaluation set:
 - epoch: 4000.0
-- eval_absolute_seg_boundary_fscore_Boundary Prediction: 0.5797
-- eval_absolute_seg_boundary_fscore_Entropy: 0.5438
-- eval_absolute_seg_boundary_fscore_Increase in Boundary Prediction: 0.6442
-- eval_absolute_seg_boundary_fscore_Increase in Entropy: 0.6066
-- eval_absolute_seg_boundary_fscore_Increase in Loss: 0.6391
-- eval_absolute_seg_boundary_fscore_Increase in Rank: 0.6575
-- eval_absolute_seg_boundary_fscore_Loss: 0.5624
-- eval_absolute_seg_boundary_fscore_Majority Vote Cutoff: 0.6868
-- eval_absolute_seg_boundary_fscore_Majority Vote Spike: 0.6908
-- eval_absolute_seg_boundary_fscore_Rank: 0.5692
-- eval_absolute_seg_type_fscore_Boundary Prediction: 0.1765
-- eval_absolute_seg_type_fscore_Entropy: 0.2690
-- eval_absolute_seg_type_fscore_Increase in Boundary Prediction: 0.1667
-- eval_absolute_seg_type_fscore_Increase in Entropy: 0.3180
-- eval_absolute_seg_type_fscore_Increase in Loss: 0.3435
-- eval_absolute_seg_type_fscore_Increase in Rank: 0.3938
-- eval_absolute_seg_type_fscore_Loss: 0.2270
-- eval_absolute_seg_type_fscore_Majority Vote Cutoff: 0.3964
-- eval_absolute_seg_type_fscore_Majority Vote Spike: 0.4137
-- eval_absolute_seg_type_fscore_Rank: 0.2869
-- eval_bpc: 4.3835
-- eval_loss: 3.0384
+- eval_absolute_seg_boundary_fscore_Boundary Prediction: 0.5470
+- eval_absolute_seg_boundary_fscore_Entropy: 0.4522
+- eval_absolute_seg_boundary_fscore_Increase in Boundary Prediction: 0.5803
+- eval_absolute_seg_boundary_fscore_Increase in Entropy: 0.4833
+- eval_absolute_seg_boundary_fscore_Increase in Loss: 0.5784
+- eval_absolute_seg_boundary_fscore_Increase in Rank: 0.6139
+- eval_absolute_seg_boundary_fscore_Loss: 0.4960
+- eval_absolute_seg_boundary_fscore_Majority Vote Cutoff: 0.6259
+- eval_absolute_seg_boundary_fscore_Majority Vote Spike: 0.6291
+- eval_absolute_seg_boundary_fscore_Rank: 0.5341
+- eval_absolute_seg_type_fscore_Boundary Prediction: 0.3005
+- eval_absolute_seg_type_fscore_Entropy: 0.2714
+- eval_absolute_seg_type_fscore_Increase in Boundary Prediction: 0.3422
+- eval_absolute_seg_type_fscore_Increase in Entropy: 0.2764
+- eval_absolute_seg_type_fscore_Increase in Loss: 0.3523
+- eval_absolute_seg_type_fscore_Increase in Rank: 0.3931
+- eval_absolute_seg_type_fscore_Loss: 0.2706
+- eval_absolute_seg_type_fscore_Majority Vote Cutoff: 0.4061
+- eval_absolute_seg_type_fscore_Majority Vote Spike: 0.3590
+- eval_absolute_seg_type_fscore_Rank: 0.2985
+- eval_bpc: 4.5128
+- eval_loss: 3.1280
 - eval_model_preparation_time: 0.0008
-- eval_perplexity: 20.8724
-- eval_runtime: 7.5077
-- eval_samples_per_second: 21.711
-- eval_spike_seg_boundary_fscore_Boundary Prediction: 0.6793
-- eval_spike_seg_boundary_fscore_Entropy: 0.5896
-- eval_spike_seg_boundary_fscore_Increase in Boundary Prediction: 0.6781
-- eval_spike_seg_boundary_fscore_Increase in Entropy: 0.6219
-- eval_spike_seg_boundary_fscore_Increase in Loss: 0.6493
-- eval_spike_seg_boundary_fscore_Increase in Rank: 0.6472
-- eval_spike_seg_boundary_fscore_Loss: 0.5906
-- eval_spike_seg_boundary_fscore_Majority Vote Cutoff: 0.7049
-- eval_spike_seg_boundary_fscore_Majority Vote Spike: 0.6959
-- eval_spike_seg_boundary_fscore_Rank: 0.5967
-- eval_spike_seg_type_fscore_Boundary Prediction: 0.344
-- eval_spike_seg_type_fscore_Entropy: 0.2832
-- eval_spike_seg_type_fscore_Increase in Boundary Prediction: 0.3373
-- eval_spike_seg_type_fscore_Increase in Entropy: 0.2921
-- eval_spike_seg_type_fscore_Increase in Loss: 0.3162
-- eval_spike_seg_type_fscore_Increase in Rank: 0.3674
-- eval_spike_seg_type_fscore_Loss: 0.2706
-- eval_spike_seg_type_fscore_Majority Vote Cutoff: 0.3896
-- eval_spike_seg_type_fscore_Majority Vote Spike: 0.3279
-- eval_spike_seg_type_fscore_Rank: 0.3246
-- eval_steps_per_second: 0.799
+- eval_perplexity: 22.8288
+- eval_runtime: 12.2624
+- eval_samples_per_second: 12.151
+- eval_spike_seg_boundary_fscore_Boundary Prediction: 0.5811
+- eval_spike_seg_boundary_fscore_Entropy: 0.4902
+- eval_spike_seg_boundary_fscore_Increase in Boundary Prediction: 0.5768
+- eval_spike_seg_boundary_fscore_Increase in Entropy: 0.4836
+- eval_spike_seg_boundary_fscore_Increase in Loss: 0.5501
+- eval_spike_seg_boundary_fscore_Increase in Rank: 0.5845
+- eval_spike_seg_boundary_fscore_Loss: 0.5259
+- eval_spike_seg_boundary_fscore_Majority Vote Cutoff: 0.6380
+- eval_spike_seg_boundary_fscore_Majority Vote Spike: 0.6029
+- eval_spike_seg_boundary_fscore_Rank: 0.5890
+- eval_spike_seg_type_fscore_Boundary Prediction: 0.2831
+- eval_spike_seg_type_fscore_Entropy: 0.2583
+- eval_spike_seg_type_fscore_Increase in Boundary Prediction: 0.2799
+- eval_spike_seg_type_fscore_Increase in Entropy: 0.2175
+- eval_spike_seg_type_fscore_Increase in Loss: 0.2712
+- eval_spike_seg_type_fscore_Increase in Rank: 0.3042
+- eval_spike_seg_type_fscore_Loss: 0.2594
+- eval_spike_seg_type_fscore_Majority Vote Cutoff: 0.3492
+- eval_spike_seg_type_fscore_Majority Vote Spike: 0.2747
+- eval_spike_seg_type_fscore_Rank: 0.3375
+- eval_steps_per_second: 0.408
 - step: 100000
 
 ## Model description

config.json

@@ -27,5 +27,5 @@
   "torch_dtype": "float32",
   "transformers_version": "4.44.2",
   "use_cache": true,
-  "vocab_size": 48
+  "vocab_size": 34
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:78f6926d87d45b54c7fa5fdcbca74dd73ac3ef421ef2beeb587acfa2ecb88b76
-size 2539800
+oid sha256:efe533efbd0b31c095b2145cb7c348667d0ec3699e849cc3981aa0d82694b3c0
+size 2532632

tokenizer.json

@@ -115,50 +115,36 @@
       "PAD": 1,
       "WORD_BOUNDARY": 2,
       "UTT_BOUNDARY": 3,
-      "ɔ": 4,
-      "h": 5,
-      "m": 6,
-      "ai": 7,
-      "ɛ": 8,
-      "r": 9,
-      "t": 10,
-      "ɑː": 11,
-      "p": 12,
-      "d": 13,
-      "iː": 14,
-      "b": 15,
-      "oː": 16,
-      "f": 17,
-      "eː": 18,
-      "χ": 19,
-      "w": 20,
-      "a": 21,
-      "n": 22,
-      "ø": 23,
-      "j": 24,
-      "au": 25,
-      "ə": 26,
-      "ɔi": 27,
-      "ð": 28,
-      "ɪ": 29,
-      "s": 30,
-      "ɡ": 31,
-      "ʊi": 32,
-      "ʊ": 33,
-      "əi": 34,
-      "θ": 35,
-      "l": 36,
-      "ʌ": 37,
-      "ŋ": 38,
-      "v": 39,
-      "k": 40,
-      "ɬ": 41,
-      "ɪu": 42,
-      "uː": 43,
-      "ʃ": 44,
-      "ɛu": 45,
-      "d̠ʒ": 46,
-      "z": 47
+      "j": 4,
+      "e̞": 5,
+      "s̪̻": 6,
+      "t̪̻": 7,
+      "u": 8,
+      "l": 9,
+      "o̞": 10,
+      "ʒ̺": 11,
+      "i": 12,
+      "ʋ": 13,
+      "d̪̻": 14,
+      "ä": 15,
+      "m": 16,
+      "n": 17,
+      "r": 18,
+      "k": 19,
+      "t̪̻s̪̻": 20,
+      "p": 21,
+      "ʃ̺": 22,
+      "x": 23,
+      "b": 24,
+      "ɡ": 25,
+      "t̻ʃ̻": 26,
+      "f": 27,
+      "z̪̻": 28,
+      "ɲ": 29,
+      "ʎ": 30,
+      "d̻ʒ̻": 31,
+      "y": 32,
+      "w": 33
     },
     "unk_token": "UNK"
   }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:193dccba324f20834b381484ac1135bae01184adac5a817a5c4fd3bb127ce078
+oid sha256:52482822ae0aead84723e50320eafa35fe04b4377ff02a99470964996e89b1f0
 size 5368

vocab.json

@@ -1 +1 @@
-{"UNK":0,"PAD":1,"WORD_BOUNDARY":2,"UTT_BOUNDARY":3,"ɔ":4,"h":5,"m":6,"ai":7,"ɛ":8,"r":9,"t":10,"ɑː":11,"p":12,"d":13,"iː":14,"b":15,"oː":16,"f":17,"eː":18,"χ":19,"w":20,"a":21,"n":22,"ø":23,"j":24,"au":25,"ə":26,"ɔi":27,"ð":28,"ɪ":29,"s":30,"ɡ":31,"ʊi":32,"ʊ":33,"əi":34,"θ":35,"l":36,"ʌ":37,"ŋ":38,"v":39,"k":40,"ɬ":41,"ɪu":42,"uː":43,"ʃ":44,"ɛu":45,"d̠ʒ":46,"z":47}
+{"UNK":0,"PAD":1,"WORD_BOUNDARY":2,"UTT_BOUNDARY":3,"j":4,"e̞":5,"s̪̻":6,"t̪̻":7,"u":8,"l":9,"o̞":10,"ʒ̺":11,"i":12,"ʋ":13,"d̪̻":14,"ä":15,"m":16,"n":17,"r":18,"k":19,"t̪̻s̪̻":20,"p":21,"ʃ̺":22,"x":23,"b":24,"ɡ":25,"t̻ʃ̻":26,"f":27,"z̪̻":28,"ɲ":29,"ʎ":30,"d̻ʒ̻":31,"y":32,"w":33}