chore: first model is now ready

Author: jortsupetterson

models/best_models/eng/best.pt

@@ -1,7 +1,7 @@
 {
-  "epoch": 9,
-  "train_loss": 1.9133409197266038,
-  "validation_loss": 2.0183915266623864,
+  "epoch": 30,
+  "train_loss": 0.805634579261144,
+  "validation_loss": 1.290153529047966,
   "train_exact_match": null,
   "validation_exact_match": null
 }

models/best_models/eng/metrics.json

@@ -2,23 +2,23 @@
   "language": "eng",
   "corpusPath": "C:\\Users\\jorts\\order\\models\\tokenizers\\eng\\corpus.jsonl",
   "modelPath": "C:\\Users\\jorts\\order\\models\\tokenizers\\eng\\tokenizer.model",
-  "sampleCount": 57,
-  "trainCount": 43,
-  "validationCount": 14,
+  "sampleCount": 100,
+  "trainCount": 75,
+  "validationCount": 25,
   "inputLengths": {
-    "count": 57,
-    "min": 34,
-    "max": 422,
-    "avg": 198.47368421052633,
-    "p50": 191,
-    "p95": 334
+    "count": 100,
+    "min": 27,
+    "max": 480,
+    "avg": 184.92,
+    "p50": 169,
+    "p95": 376
   },
   "labelLengths": {
-    "count": 57,
-    "min": 127,
-    "max": 1899,
-    "avg": 655.6842105263158,
-    "p50": 627,
-    "p95": 1162
+    "count": 100,
+    "min": 59,
+    "max": 1849,
+    "avg": 601.2,
+    "p50": 609,
+    "p95": 1036
   }
 }

models/datasets/eng/stats.json

@@ -1,23 +1,29 @@
 {
   "language": "eng",
-  "format": "onnx-seq2seq-distribution",
+  "format": "onnx-seq2seq-build",
   "modelFile": "model.onnx",
   "tokenizerModelFile": "tokenizer.model",
   "tokenizerVocabFile": "tokenizer.vocab",
   "metricsFile": "metrics.json",
-  "inputNames": ["input_ids", "attention_mask", "decoder_input_ids"],
-  "outputNames": ["logits"],
+  "inputNames": [
+    "input_ids",
+    "attention_mask",
+    "decoder_input_ids"
+  ],
+  "outputNames": [
+    "logits"
+  ],
   "tokenIds": {
     "bos": 1,
     "eos": 2,
-    "pad": 665
+    "pad": 4329
   },
   "limits": {
-    "maxInputLength": 329,
-    "maxDecoderLength": 1329
+    "maxInputLength": 256,
+    "maxDecoderLength": 513
   },
   "model": {
-    "vocabSize": 665,
+    "vocabSize": 4329,
     "dModel": 128,
     "numHeads": 4,
     "numEncoderLayers": 2,
@@ -27,11 +33,24 @@
   },
   "validation": {
     "dummy_input_shape": {
-      "input_ids": [1, 16],
-      "attention_mask": [1, 16],
-      "decoder_input_ids": [1, 8]
+      "input_ids": [
+        1,
+        16
+      ],
+      "attention_mask": [
+        1,
+        16
+      ],
+      "decoder_input_ids": [
+        1,
+        8
+      ]
     },
-    "output_shape": [1, 8, 665],
-    "max_abs_diff": 9.5367431640625e-7
+    "output_shape": [
+      1,
+      8,
+      4329
+    ],
+    "max_abs_diff": 4.76837158203125e-06
   }
 }

models/datasets/eng/train.jsonl

@@ -1,7 +1,7 @@
 {
-  "epoch": 1,
-  "train_loss": 6.571187973022461,
-  "validation_loss": 6.362093925476074,
+  "epoch": 30,
+  "train_loss": 0.805634579261144,
+  "validation_loss": 1.290153529047966,
   "train_exact_match": null,
   "validation_exact_match": null
 }