tmp-spec-checkpoint-135000/config.json

{
  "attention_bias": false,
  "attention_dropout": 0.0,
  "bos_token_id": 2,
  "eos_token_id": 3,
  "fused_bias_fc": false,
  "fused_dropout_add_ln": false,
  "fused_mlp": false,
  "head_dim": 64,
  "hidden_act": "silu",
  "hidden_size": 640,
  "initializer_range": 0.02,
  "intermediate_size": 2560,
  "loss_type": "ForCausalLM",
  "max_position_embeddings": 2048,
  "max_seq_length": 64,
  "mlp_bias": false,
  "model_type": "llama",
  "num_attention_heads": 10,
  "num_hidden_layers": 24,
  "num_key_value_heads": 10,
  "pretraining_tp": 1,
  "residual_in_fp32": true,
  "rms_norm_eps": 1e-06,
  "rope_scaling": null,
  "rope_theta": 10000.0,
  "tie_word_embeddings": false,
  "transformers_version": "4.46.2",
  "use_cache": true,
  "use_flash_attn": true,
  "vocab_size": 84
}