{
  "model_name": "hermes-mobile-270m-litertlm",
  "display_name": "Hermes Mobile 270M (LiteRT-LM) — iPhone 16 optimized",
  "format": "litertlm",
  "quantization": "int4_per_channel",
  "context_length": 4096,
  "parameters": "270M",
  "framework": "litert-lm",
  "architecture": "decoder-only-gqa",
  "license": "apache-2.0",
  "tags": [
    "litert-lm",
    "litert",
    "on-device",
    "mobile",
    "iphone",
    "ios",
    "iphone-16",
    "apple-neural-engine",
    "agent",
    "tool-calling",
    "google-ai-edge-gallery",
    "raven-ecosystem"
  ],
  "library_name": "litert-lm",
  "pipeline_tag": "text-generation",
  "base_model": "hermes-edge/hermes-270m",
  "intended_org": "bclermo",
  "files": {
    "model": "hermes-mobile-270m-int4.litertlm",
    "model_card": "model_card.md"
  },
  "config": {
    "vocab_size": 32000,
    "hidden_size": 1024,
    "intermediate_size": 2816,
    "num_layers": 21,
    "num_heads": 16,
    "num_kv_heads": 4,
    "head_dim": 64,
    "max_seq_len": 4096,
    "rope_theta": 10000.0,
    "rms_norm_eps": 1e-06,
    "tie_embeddings": true,
    "attention": "grouped_query_attention",
    "activation": "silu_swiglu",
    "normalization": "rms_norm",
    "position_embedding": "rope"
  },
  "runtime": {
    "min_ios_version": 17,
    "recommended_ios_version": 18,
    "min_android_version": 10,
    "recommended_ram_gb": 6,
    "accelerators": ["apple_neural_engine", "gpu", "cpu"],
    "apple_ane": true,
    "backends": ["coreml", "gpu", "cpu"],
    "approx_file_size_mb": 180
  },
  "variants": {
    "hermes-270m": {"size_mb": 180, "params": "270M", "iphone_16_tok_per_sec": 55},
    "hermes-500m": {"size_mb": 280, "params": "500M", "iphone_16_tok_per_sec": 40},
    "hermes-1b": {"size_mb": 600, "params": "1B", "iphone_16_tok_per_sec": 25}
  }
}