config.json

{
  "a_ps": [
    [
      "dp",
      "fsdp"
    ],
    "sp",
    "tp",
    null
  ],
  "architectures": [
    "LlamaForCausalLM"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "axis_dims": [
    1,
    -1,
    1,
    1
  ],
  "axis_names": [
    "dp",
    "fsdp",
    "tp",
    "sp"
  ],
  "b_ps": [
    [
      "dp",
      "fsdp"
    ],
    null,
    null,
    null
  ],
  "backend": null,
  "bits": null,
  "bos_token_id": 1,
  "c_max_position_embeddings": 2048,
  "easy_method": "train",
  "embd_pdrop": 0.0,
  "eos_token_id": 2,
  "fcm_max_ratio": 0.0,
  "fcm_min_ratio": 0.0,
  "flash_attn_key_chunk_size": 1024,
  "flash_attn_query_chunk_size": 1024,
  "freq_max_position_embeddings": 2048,
  "hidden_act": "silu",
  "hidden_size": 2048,
  "initializer_range": 0.02,
  "intermediate_size": 5632,
  "k_ps": [
    [
      "dp",
      "fsdp"
    ],
    "sp",
    "tp",
    null
  ],
  "max_position_embeddings": 2048,
  "model_type": "llama",
  "num_attention_heads": 32,
  "num_hidden_layers": 22,
  "num_key_value_heads": 4,
  "number_rep_kv": 1,
  "pretraining_tp": 1,
  "q_ps": [
    [
      "dp",
      "fsdp"
    ],
    "sp",
    "tp",
    null
  ],
  "resid_pdrop": 0.0,
  "rms_norm_eps": 1e-05,
  "rope_scaling": null,
  "rope_theta": 10000.0,
  "scan_layers": true,
  "scan_mlp_chunk_size": 1024,
  "tie_word_embeddings": false,
  "torch_dtype": "float16",
  "transformers_version": "4.36.2",
  "use_cache": true,
  "use_flash_attention": false,
  "use_pjit_attention_force": false,
  "use_sacn_mlp": false,
  "use_shard_map": false,
  "v_ps": [
    [
      "dp",
      "fsdp"
    ],
    "sp",
    "tp",
    null
  ],
  "vocab_size": 32000
}