{
  "id": "code-llama-7b",
  "name": "Code Llama 7B",
  "identifiers": {
    "huggingface": [
      "meta-llama/CodeLlama-7b-hf",
      "codellama/CodeLlama-7b-hf"
    ]
  },
  "developer": "Meta",
  "release_date": {
    "value": "2023-08-24",
    "source": "https://arxiv.org/abs/2308.12950",
    "status": "recorded"
  },
  "weights_status": "open",
  "availability": {
    "value": "gated",
    "checked": "2026-09-24",
    "source": "https://huggingface.co/meta-llama/CodeLlama-7b-hf",
    "note": "Meta's repo is gated (manual approval); the codellama/ transformers conversion used for the config is not gated."
  },
  "license": {
    "value": "Released under the same custom license as Llama 2 (paper: 'released under the same custom permissive license').",
    "source": "https://arxiv.org/abs/2308.12950",
    "status": "recorded"
  },
  "architecture": {
    "family": "decoder_only",
    "n_layers": {
      "value": 32,
      "source": "https://huggingface.co/codellama/CodeLlama-7b-hf/raw/6c284d1468fe6c413cf56183e69b194dcfa27fe6/config.json",
      "status": "recorded"
    },
    "hidden_size": {
      "value": 4096,
      "source": "https://huggingface.co/codellama/CodeLlama-7b-hf/raw/6c284d1468fe6c413cf56183e69b194dcfa27fe6/config.json",
      "status": "recorded"
    },
    "n_heads": {
      "value": 32,
      "source": "https://huggingface.co/codellama/CodeLlama-7b-hf/raw/6c284d1468fe6c413cf56183e69b194dcfa27fe6/config.json",
      "status": "recorded"
    },
    "vocab_size": {
      "value": 32016,
      "source": "https://huggingface.co/codellama/CodeLlama-7b-hf/raw/6c284d1468fe6c413cf56183e69b194dcfa27fe6/config.json",
      "status": "recorded"
    },
    "positional_encoding": {
      "value": "rotary (RoPE)",
      "source": "https://huggingface.co/codellama/CodeLlama-7b-hf/raw/6c284d1468fe6c413cf56183e69b194dcfa27fe6/config.json",
      "status": "recorded"
    },
    "context_length": {
      "value": 16384,
      "source": "https://huggingface.co/codellama/CodeLlama-7b-hf/raw/6c284d1468fe6c413cf56183e69b194dcfa27fe6/config.json",
      "status": "recorded"
    },
    "n_kv_heads": {
      "value": 32,
      "source": "https://huggingface.co/codellama/CodeLlama-7b-hf/raw/6c284d1468fe6c413cf56183e69b194dcfa27fe6/config.json",
      "status": "recorded"
    },
    "note": "Config is from the codellama/ HF org's transformers conversion, pinned by commit. Paper: RoPE base period raised from 10,000 to 1,000,000 for long-context fine-tuning (config rope_theta = 1000000). Vocab 32016 vs Llama 2's 32000; the extra tokens are not explained in the sources read."
  },
  "training_data": {
    "value": "Initialized from Llama 2 7B, then 500B tokens of 'a near-deduplicated dataset of publicly available code' (8% natural language about code).",
    "source": "https://arxiv.org/abs/2308.12950",
    "status": "partial"
  },
  "techniques": [
    "transformer-decoder",
    "rotary-position-embedding",
    "rmsnorm",
    "swiglu"
  ],
  "primary_sources": [
    "https://arxiv.org/abs/2308.12950",
    "https://huggingface.co/codellama/CodeLlama-7b-hf/raw/6c284d1468fe6c413cf56183e69b194dcfa27fe6/config.json"
  ],
  "record_history": [
    {
      "date": "2026-09-24",
      "change": "created from primary sources (Phase 1 seed, batch 1)",
      "by": "wilson-pruitt + claude"
    },
    {
      "date": "2026-09-24",
      "change": "availability checked and recorded",
      "by": "wilson-pruitt + claude"
    }
  ]
}