CoreMLModels/meta.json

{
  "context_length": 4096,
  "prefill_length": 256,
  "batch_size": 1,
  "hidden_size": 2048,
  "vocab_size": 65536,
  "num_attention_heads": 32,
  "num_key_value_heads": 8,
  "head_dim": 64,
  "conv_kernel_size": 3,
  "bos_token_id": 1,
  "eos_token_id": 7,
  "pad_token_id": 0,
  "image_token_id": 396,
  "image_start_token_id": 498,
  "image_end_token_id": 499,
  "image_thumbnail_token_id": 497,
  "tile_size": 512,
  "encoder_patch_size": 16,
  "downsample_factor": 2,
  "min_tiles": 2,
  "max_tiles": 10,
  "use_thumbnail": true,
  "min_image_tokens": 64,
  "max_image_tokens": 256,
  "max_num_patches": 1024,
  "vision_hidden_size": 1152,
  "projector_hidden_size": 2048,
  "vision_position_embedding_size": 16,
  "vision_patch_embedding_model": "VisionPatchEmbedding.mlpackage",
  "vision_encoder_model": "VisionEncoder.mlpackage",
  "vision_projector_model": "VisionProjector.mlpackage",
  "vision_position_embeddings_file": "VisionPositionEmbeddings.float16.bin",
  "chunk_order": [
    "DecoderChunk0",
    "DecoderChunk1",
    "DecoderChunk2",
    "DecoderChunk3"
  ],
  "chunks": [
    {
      "name": "DecoderChunk0",
      "start_layer": 0,
      "end_layer": 4,
      "conv_layer_indices": [
        0,
        1,
        3
      ],
      "attention_layer_indices": [
        2
      ]
    },
    {
      "name": "DecoderChunk1",
      "start_layer": 4,
      "end_layer": 8,
      "conv_layer_indices": [
        4,
        6,
        7
      ],
      "attention_layer_indices": [
        5
      ]
    },
    {
      "name": "DecoderChunk2",
      "start_layer": 8,
      "end_layer": 12,
      "conv_layer_indices": [
        9,
        11
      ],
      "attention_layer_indices": [
        8,
        10
      ]
    },
    {
      "name": "DecoderChunk3",
      "start_layer": 12,
      "end_layer": 16,
      "conv_layer_indices": [
        13,
        15
      ],
      "attention_layer_indices": [
        12,
        14
      ]
    }
  ],
  "layer_types": [
    "conv",
    "conv",
    "full_attention",
    "conv",
    "conv",
    "full_attention",
    "conv",
    "conv",
    "full_attention",
    "conv",
    "full_attention",
    "conv",
    "full_attention",
    "conv",
    "full_attention",
    "conv"
  ],
  "embeddings_model": "Embeddings.mlpackage",
  "lm_head_model": "LMHead.mlpackage",
  "prefill_models": [
    "DecoderChunk0Prefill.mlpackage",
    "DecoderChunk1Prefill.mlpackage",
    "DecoderChunk2Prefill.mlpackage",
    "DecoderChunk3Prefill.mlpackage"
  ],
  "decode_models": [
    "DecoderChunk0Decode.mlpackage",
    "DecoderChunk1Decode.mlpackage",
    "DecoderChunk2Decode.mlpackage",
    "DecoderChunk3Decode.mlpackage"
  ],
  "chat_template": "{{- bos_token -}}\n{%- set keep_past_thinking = keep_past_thinking | default(false) -%}\n{%- set ns = namespace(system_prompt=\"\") -%}\n{%- if messages[0][\"role\"] == \"system\" -%}\n    {%- set sys_content = messages[0][\"content\"] -%}\n    {%- if sys_content is not string -%}\n        {%- for item in sys_content -%}\n            {%- if item[\"type\"] == \"text\" -%}\n                {%- set ns.system_prompt = ns.system_prompt + item[\"text\"] -%}\n            {%- endif -%}\n        {%- endfor -%}\n    {%- else -%}\n        {%- set ns.system_prompt = sys_content -%}\n    {%- endif -%}\n    {%- set messages = messages[1:] -%}\n{%- endif -%}\n{%- if tools -%}\n    {%- set ns.system_prompt = ns.system_prompt + (\"\\n\" if ns.system_prompt else \"\") + \"List of tools: [\" -%}\n    {%- for tool in tools -%}\n        {%- if tool is not string -%}\n            {%- set tool = tool | tojson -%}\n        {%- endif -%}\n        {%- set ns.system_prompt = ns.system_prompt + tool -%}\n        {%- if not loop.last -%}\n            {%- set ns.system_prompt = ns.system_prompt + \", \" -%}\n        {%- endif -%}\n    {%- endfor -%}\n    {%- set ns.system_prompt = ns.system_prompt + \"]\" -%}\n{%- endif -%}\n{%- if ns.system_prompt -%}\n    {{- \"<|im_start|>system\\n\" + ns.system_prompt + \"<|im_end|>\\n\" -}}\n{%- endif -%}\n{%- set ns.last_assistant_index = -1 -%}\n{%- for message in messages -%}\n    {%- if message[\"role\"] == \"assistant\" -%}\n        {%- set ns.last_assistant_index = loop.index0 -%}\n    {%- endif -%}\n{%- endfor -%}\n{%- for message in messages -%}\n    {{- \"<|im_start|>\" + message[\"role\"] + \"\\n\" -}}\n    {%- if message[\"content\"] is not string -%}\n        {%- set ns.content = \"\" -%}\n        {%- for item in message[\"content\"] -%}\n            {%- if item[\"type\"] == \"image\" -%}\n                {%- set ns.content = ns.content + \"<image>\" -%}\n            {%- elif item[\"type\"] == \"text\" -%}\n                {%- set ns.content = ns.content + item[\"text\"] -%}\n            {%- else -%}\n                {%- set ns.content = ns.content + item | tojson -%}\n            {%- endif -%}\n        {%- endfor -%}\n        {%- set content = ns.content -%}\n    {%- else -%}\n        {%- set content = message[\"content\"] -%}\n    {%- endif -%}\n    {%- if message[\"role\"] == \"assistant\" and not keep_past_thinking and loop.index0 != ns.last_assistant_index -%}\n        {%- if \"</think>\" in content -%}\n            {%- set content = content.split(\"</think>\")[-1] | trim -%}\n        {%- endif -%}\n    {%- endif -%}\n    {{- content + \"<|im_end|>\\n\" -}}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n    {{- \"<|im_start|>assistant\\n\" -}}\n{%- endif -%}"
}