Some more changes to split files per program

HuggingFace_Mistral_Transformer_Single_Instrument.ipynb

@@ -85,19 +85,19 @@
     "# Our tokenizer's configuration\n",
     "BEAT_RES = {(0, 1): 12, (1, 2): 4, (2, 4): 2, (4, 8): 1}\n",
     "TOKENIZER_PARAMS = {\n",
-    "    \"pitch_range\": (21, 109),\n",
+    "    \"pitch_range\": (21, 108),\n",
     "    \"beat_res\": BEAT_RES,\n",
-    "    \"num_velocities\": 24,\n",
+    "    \"num_velocities\": 32,\n",
     "    \"special_tokens\": [\"PAD\", \"BOS\", \"EOS\"],\n",
     "    \"use_chords\": True,\n",
     "    \"use_rests\": True,\n",
     "    \"use_tempos\": True,\n",
     "    \"use_time_signatures\": True,\n",
     "    \"use_programs\": False,  # We want single track \n",
-    "    \"one_token_stream_for_programs\": True,\n",
+    "    \"one_token_stream_for_programs\": False, # We want single track\n",
     "    \"programs\": list(range(0, 128)), #-1 drums, skip drums\n",
     "    \"num_tempos\": 32,\n",
-    "    \"tempo_range\": (50, 200),  # (min_tempo, max_tempo)\n",
+    "    \"tempo_range\": (40, 250),  # (min_tempo, max_tempo)\n",
     "}\n",
     "config = TokenizerConfig(**TOKENIZER_PARAMS)\n",
     "\n",
@@ -485,7 +485,7 @@
     "    \n",
     ")\n",
     "\n",
-    "collator = DataCollator(tokenizer[\"PAD_None\"], copy_inputs_as_labels=True)\n",
+    "collator = DataCollator(tokenizer[\"PAD_None\"], copy_inputs_as_labels=True, pad_on_left=True) #not sure about the pad_on_left, it might get better results\n",
     "trainer = Trainer(\n",
     "    model=model,\n",
     "    args=training_config,\n",