Training in progress, step 12

adapter_config.json

@@ -1,10 +1,10 @@
 {
   "alpha_pattern": {},
   "auto_mapping": {
-    "base_model_class": "MusicgenMelodyForConditionalGeneration",
-    "parent_library": "transformers.models.musicgen_melody.modeling_musicgen_melody"
+    "base_model_class": "MusicgenForConditionalGeneration",
+    "parent_library": "transformers.models.musicgen.modeling_musicgen"
   },
-  "base_model_name_or_path": "facebook/musicgen-melody",
+  "base_model_name_or_path": "facebook/musicgen-small",
   "bias": "none",
   "fan_in_fan_out": false,
   "inference_mode": true,
@@ -24,21 +24,21 @@
   "revision": null,
   "target_modules": [
     "embed_tokens.0",
-    "lm_heads.0",
-    "embed_tokens.3",
-    "fc1",
+    "embed_tokens.2",
+    "enc_to_dec_proj",
     "q_proj",
-    "embed_tokens.1",
-    "k_proj",
     "lm_heads.2",
-    "enc_to_dec_proj",
-    "v_proj",
     "audio_enc_to_dec_proj",
+    "v_proj",
     "out_proj",
-    "lm_heads.3",
-    "embed_tokens.2",
+    "fc1",
+    "lm_heads.0",
+    "k_proj",
     "lm_heads.1",
-    "fc2"
+    "fc2",
+    "embed_tokens.1",
+    "embed_tokens.3",
+    "lm_heads.3"
   ],
   "task_type": null,
   "use_dora": false,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f2c0528c768e1f85befa94bed793612e14262ad71afd4d59a7c4fdd35b4ba3a5
-size 87103456
+oid sha256:82f5e1ae6bb4d18651b8f9dd23a4aa60a2063a4ba73867bb7abfb54939ad6c79
+size 42656200

config.json

@@ -1,7 +1,7 @@
 {
-  "_name_or_path": "facebook/musicgen-melody",
+  "_name_or_path": "facebook/musicgen-small",
   "architectures": [
-    "MusicgenMelodyForConditionalGeneration"
+    "MusicgenForConditionalGeneration"
   ],
   "audio_encoder": {
     "_attn_implementation_autoset": false,
@@ -99,7 +99,6 @@
     "use_causal_conv": false,
     "use_conv_shortcut": false
   },
-  "chroma_length": 235,
   "decoder": {
     "_attn_implementation_autoset": false,
     "_name_or_path": "",
@@ -113,6 +112,7 @@
     "begin_suppress_tokens": null,
     "bos_token_id": 2048,
     "chunk_size_feed_forward": 0,
+    "classifier_dropout": 0.0,
     "cross_attention_hidden_size": null,
     "decoder_start_token_id": 2048,
     "diversity_penalty": 0.0,
@@ -122,11 +122,11 @@
     "encoder_no_repeat_ngram_size": 0,
     "eos_token_id": null,
     "exponential_decay_length_penalty": null,
-    "ffn_dim": 6144,
+    "ffn_dim": 4096,
     "finetuning_task": null,
     "forced_bos_token_id": null,
     "forced_eos_token_id": null,
-    "hidden_size": 1536,
+    "hidden_size": 1024,
     "id2label": {
       "0": "LABEL_0",
       "1": "LABEL_1"
@@ -143,13 +143,13 @@
     "max_length": 20,
     "max_position_embeddings": 2048,
     "min_length": 0,
-    "model_type": "musicgen_melody_decoder",
+    "model_type": "musicgen_decoder",
     "no_repeat_ngram_size": 0,
-    "num_attention_heads": 24,
+    "num_attention_heads": 16,
     "num_beam_groups": 1,
     "num_beams": 1,
     "num_codebooks": 4,
-    "num_hidden_layers": 48,
+    "num_hidden_layers": 24,
     "num_return_sequences": 1,
     "output_attentions": false,
     "output_hidden_states": false,
@@ -181,8 +181,8 @@
     "vocab_size": 2048
   },
   "decoder_start_token_id": 2048,
-  "model_type": "musicgen_melody",
-  "num_chroma": 12,
+  "is_encoder_decoder": true,
+  "model_type": "musicgen",
   "pad_token_id": 2048,
   "text_encoder": {
     "_attn_implementation_autoset": false,

preprocessor_config.json

@@ -1,17 +1,11 @@
 {
-  "chunk_length": 30,
-  "feature_extractor_type": "MusicgenMelodyFeatureExtractor",
-  "feature_size": 12,
-  "hop_length": 4096,
-  "n_fft": 16384,
-  "n_samples": 960000,
-  "padding_side": "right",
+  "chunk_length_s": null,
+  "feature_extractor_type": "EncodecFeatureExtractor",
+  "feature_size": 1,
+  "overlap": null,
+  "padding_side": "left",
   "padding_value": 0.0,
-  "processor_class": "MusicgenMelodyProcessor",
-  "return_attention_mask": false,
-  "sampling_rate": 32000,
-  "stem_indices": [
-    3,
-    2
-  ]
+  "processor_class": "MusicgenProcessor",
+  "return_attention_mask": true,
+  "sampling_rate": 32000
 }

tokenizer_config.json

@@ -931,13 +931,9 @@
   "eos_token": "</s>",
   "extra_ids": 100,
   "extra_special_tokens": {},
-  "max_length": null,
   "model_max_length": 512,
-  "pad_to_multiple_of": null,
   "pad_token": "<pad>",
-  "pad_token_type_id": 0,
-  "padding_side": "right",
-  "processor_class": "MusicgenMelodyProcessor",
+  "processor_class": "MusicgenProcessor",
   "tokenizer_class": "T5Tokenizer",
   "unk_token": "<unk>"
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:41809a74fb3412207e0b4b4026dd4bdaec2d1456df555149b425b82fafea7953
+oid sha256:922234e6f345a216f8dbd9724f75753c9c265839c522389678ec2c5e0b1d3ad3
 size 5496