Update tiny models for Phi4MultimodalModel

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

chat_template.jinja

@@ -0,0 +1 @@
+{% for message in messages %}{{ '<|' + message['role'] + '|>' }}{% if message['content'] is string %}{{ message['content'] }}{% else %}{% for content in message['content'] %}{% if content['type'] == 'image' %}{{ '<|image|>' }}{% elif content['type'] == 'audio' %}{{ '<|audio|>' }}{% elif content['type'] == 'text' %}{{ content['text'] }}{% endif %}{% endfor %}{% endif %}{% if message['role'] == 'system' and 'tools' in message and message['tools'] is not none %}{{ '<|tool|>' + message['tools'] + '<|/tool|>' + '<|end|>' }}{% endif %}{{ '<|end|>' }}{% endfor %}{% if add_generation_prompt %}{{ '<|assistant|>' }}{% else %}{{ eos_token }}{% endif %}

config.json

@@ -0,0 +1,78 @@
+{
+  "architectures": [
+    "Phi4MultimodalModel"
+  ],
+  "attention_dropout": 0.0,
+  "audio_config": {
+    "activation": "swish",
+    "audio_token_id": 200011,
+    "bias_max_distance": 1000,
+    "bias_symmetric": false,
+    "chunk_size": -1,
+    "conv_activation": "swish",
+    "conv_glu_type": "swish",
+    "depthwise_multiplier": 1,
+    "depthwise_separable_out_channel": 128,
+    "downsample_rate": 1,
+    "dropout_rate": 0.0,
+    "ext_pw_out_channel": 1024,
+    "feature_layer": -2,
+    "hidden_size": 32,
+    "initializer_range": 1e-05,
+    "input_size": 80,
+    "intermediate_size": 48,
+    "kernel_size": 3,
+    "left_chunk": 18,
+    "model_type": "phi4_multimodal_audio",
+    "nemo_activation": "relu",
+    "nemo_conv_channels": 128,
+    "nemo_final_size": 10,
+    "num_attention_heads": 8,
+    "num_blocks": 2,
+    "time_reduction": 8
+  },
+  "bos_token_id": 199999,
+  "dtype": "float32",
+  "embd_pdrop": 0.0,
+  "eos_token_id": 199999,
+  "hidden_act": "silu",
+  "hidden_size": 32,
+  "initializer_range": 0.02,
+  "intermediate_size": 64,
+  "max_position_embeddings": 4096,
+  "model_type": "phi4_multimodal",
+  "num_attention_heads": 4,
+  "num_hidden_layers": 2,
+  "num_key_value_heads": 2,
+  "original_max_position_embeddings": 4096,
+  "pad_token_id": 199999,
+  "resid_pdrop": 0.0,
+  "rms_norm_eps": 1e-05,
+  "rope_parameters": {
+    "partial_rotary_factor": 1.0,
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": false,
+  "transformers_version": "5.16.0.dev0",
+  "use_cache": true,
+  "vision_config": {
+    "attention_dropout": 0.0,
+    "crop_size": 16,
+    "feature_layer": -2,
+    "hidden_act": "gelu_pytorch_tanh",
+    "hidden_size": 32,
+    "image_size": 448,
+    "image_token_id": 200010,
+    "initializer_range": 1e-05,
+    "intermediate_size": 64,
+    "layer_norm_eps": 1e-06,
+    "model_type": "phi4_multimodal_vision",
+    "num_attention_heads": 8,
+    "num_channels": 3,
+    "num_hidden_layers": 2,
+    "patch_size": 14
+  },
+  "vocab_size": 200029
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:08ffe9f4358e81cf7ca2779dca5e459609bf75e91dfa9b0edcd1e3b373ea26a5
+size 27385928

preprocessor_config.json

@@ -0,0 +1,17 @@
+{
+  "audio_compression_rate": 8,
+  "audio_downsample_rate": 1,
+  "audio_feat_stride": 1,
+  "crop_size": 448,
+  "feature_extractor_type": "Phi4MultimodalFeatureExtractor",
+  "feature_size": 80,
+  "hop_length": 160,
+  "n_fft": 512,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "preemphasis": 0.97,
+  "return_attention_mask": true,
+  "sampling_rate": 16000,
+  "size": 448,
+  "win_length": 400
+}

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aee2ac83f75c7dac7885cebefd51bbc73c6ca8176b568b9bc09383a5d7364678
+size 15524288

tokenizer_config.json

@@ -0,0 +1,21 @@
+{
+  "add_prefix_space": false,
+  "audio_token": "<|audio|>",
+  "backend": "tokenizers",
+  "bos_token": "<|endoftext|>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "image_token": "<|image|>",
+  "is_local": true,
+  "local_files_only": false,
+  "model_max_length": 4096,
+  "model_specific_special_tokens": {
+    "audio_token": "<|audio|>",
+    "image_token": "<|image|>"
+  },
+  "pad_token": "<|endoftext|>",
+  "processor_class": "Phi4MultimodalProcessor",
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": "<|endoftext|>"
+}