Update tiny models for KyutaiSpeechToTextForConditionalGeneration

config.json

@@ -0,0 +1,85 @@
+{
+  "architectures": [
+    "KyutaiSpeechToTextForConditionalGeneration"
+  ],
+  "attention_dropout": 0.1,
+  "audio_bos_token_id": 2048,
+  "audio_pad_token_id": 2048,
+  "bos_token_id": 1,
+  "codebook_vocab_size": 2049,
+  "codec_config": {
+    "_frame_rate": null,
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "audio_channels": 1,
+    "chunk_in_sec": null,
+    "codebook_dim": 16,
+    "codebook_size": 16,
+    "compress": 2,
+    "dilation_growth_rate": 2,
+    "head_dim": 8,
+    "hidden_act": "gelu",
+    "hidden_size": 16,
+    "initializer_range": 0.02,
+    "intermediate_size": 2048,
+    "kernel_size": 7,
+    "last_kernel_size": 3,
+    "layer_scale_initial_scale": 0.01,
+    "max_position_embeddings": 8000,
+    "model_type": "mimi",
+    "norm_eps": 1e-05,
+    "num_attention_heads": 2,
+    "num_filters": 8,
+    "num_hidden_layers": 2,
+    "num_key_value_heads": 2,
+    "num_quantizers": 8,
+    "num_residual_layers": 1,
+    "num_semantic_quantizers": 1,
+    "pad_mode": "constant",
+    "residual_kernel_size": 3,
+    "rope_parameters": {
+      "rope_theta": 10000.0,
+      "rope_type": "default"
+    },
+    "sampling_rate": 24000,
+    "sliding_window": 4,
+    "tie_word_embeddings": true,
+    "trim_right_ratio": 1.0,
+    "upsample_groups": 16,
+    "upsampling_ratios": [
+      8,
+      4
+    ],
+    "use_cache": false,
+    "use_causal_conv": true,
+    "use_conv_shortcut": false,
+    "use_streaming": false,
+    "vector_quantization_hidden_dimension": 16
+  },
+  "delay_in_tokens": 5,
+  "dtype": "float32",
+  "eos_token_id": null,
+  "ffn_dim": 38,
+  "frame_size": 64,
+  "head_dim": 8,
+  "hidden_act": "silu",
+  "hidden_size": 32,
+  "initializer_range": 0.02,
+  "max_position_embeddings": 512,
+  "model_type": "kyutai_speech_to_text",
+  "num_attention_heads": 4,
+  "num_codebooks": 8,
+  "num_hidden_layers": 2,
+  "num_key_value_heads": 4,
+  "pad_token_id": 0,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "sliding_window": 512,
+  "tie_word_embeddings": false,
+  "transformers_version": "5.16.0.dev0",
+  "use_cache": true,
+  "vocab_size": 4000
+}

generation_config.json

@@ -0,0 +1,13 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "codec__from_model_config": true,
+  "codec_output_attentions": false,
+  "codec_output_hidden_states": false,
+  "codec_use_cache": false,
+  "output_attentions": false,
+  "output_hidden_states": false,
+  "pad_token_id": 0,
+  "transformers_version": "5.16.0.dev0",
+  "use_cache": true
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c3ec580e827a92872e23a48b57ba841fe09d36d975eba38b06144af448f0672
+size 4365804

preprocessor_config.json

@@ -0,0 +1,12 @@
+{
+  "audio_delay_seconds": 2.5,
+  "audio_silence_prefix_seconds": 1.0,
+  "chunk_length_s": null,
+  "feature_extractor_type": "KyutaiSpeechToTextFeatureExtractor",
+  "feature_size": 1,
+  "overlap": null,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "return_attention_mask": true,
+  "sampling_rate": 24000
+}

tokenizer_config.json

@@ -0,0 +1,18 @@
+{
+  "backend": "tokenizers",
+  "bos_token_id": null,
+  "chat_template": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token_id": null,
+  "is_local": true,
+  "local_files_only": false,
+  "model_input_names": [
+    "input_ids",
+    "attention_mask"
+  ],
+  "model_max_length": 512,
+  "pad_token_id": null,
+  "processor_class": "KyutaiSpeechToTextProcessor",
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": "<unk>"
+}