Update tiny models for DiaForConditionalGeneration

config.json

@@ -0,0 +1,65 @@
+{
+  "architectures": [
+    "DiaForConditionalGeneration"
+  ],
+  "bos_token_id": null,
+  "decoder_config": {
+    "bos_token_id": 99,
+    "cross_head_dim": 8,
+    "cross_hidden_size": 16,
+    "cross_num_attention_heads": 2,
+    "cross_num_key_value_heads": 1,
+    "eos_token_id": 97,
+    "head_dim": 8,
+    "hidden_act": "silu",
+    "hidden_size": 32,
+    "initializer_range": 0.02,
+    "intermediate_size": 37,
+    "max_position_embeddings": 50,
+    "model_type": "dia_decoder",
+    "norm_eps": 1e-05,
+    "num_attention_heads": 2,
+    "num_channels": 3,
+    "num_hidden_layers": 2,
+    "num_key_value_heads": 1,
+    "pad_token_id": 98,
+    "rope_parameters": {
+      "rope_theta": 10000.0,
+      "rope_type": "default"
+    },
+    "use_cache": true,
+    "vocab_size": 100
+  },
+  "delay_pattern": [
+    0,
+    1,
+    2
+  ],
+  "dtype": "float32",
+  "encoder_config": {
+    "head_dim": 8,
+    "hidden_act": "silu",
+    "hidden_size": 16,
+    "initializer_range": 0.02,
+    "intermediate_size": 37,
+    "max_position_embeddings": 50,
+    "model_type": "dia_encoder",
+    "norm_eps": 1e-05,
+    "num_attention_heads": 2,
+    "num_hidden_layers": 2,
+    "num_key_value_heads": 2,
+    "rope_parameters": {
+      "rope_theta": 10000.0,
+      "rope_type": "default"
+    },
+    "vocab_size": 100
+  },
+  "eos_token_id": null,
+  "initializer_range": 0.02,
+  "is_encoder_decoder": true,
+  "model_type": "dia",
+  "norm_eps": 1e-05,
+  "pad_token_id": null,
+  "transformers_version": "5.16.0.dev0",
+  "use_cache": true
+}

generation_config.json

@@ -0,0 +1,10 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 99,
+  "eos_token_id": 97,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 98,
+  "transformers_version": "5.16.0.dev0",
+  "use_cache": true
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2cb593f454196cd03a1d5a32e3c2fe2a7504b36d3bd24af005e0c35dacfe88be
+size 162968

preprocessor_config.json

@@ -0,0 +1,12 @@
+{
+  "audio_bos_value": 1026,
+  "audio_eos_value": 1024,
+  "audio_pad_value": 1025,
+  "feature_extractor_type": "DacFeatureExtractor",
+  "feature_size": 1,
+  "hop_length": 512,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "return_attention_mask": true,
+  "sampling_rate": 16000
+}