config.json

{
  "_name_or_path": "biomap-research/xtrimopglm-3b-clm",
  "add_bias_linear": true,
  "add_qkv_bias": true,
  "apply_query_key_layer_scaling": true,
  "apply_residual_connection_post_layernorm": true,
  "architectures": [
    "xTrimoPGLMModel"
  ],
  "attention_dropout": 0.0,
  "attention_softmax_in_fp32": true,
  "auto_map": {
    "AutoConfig": "configuration_xtrimopglm.xTrimoPGLMConfig",
    "AutoModel": "modeling_xtrimopglm.xTrimoPGLMForMaskedLM",
    "AutoModelForCausalLM": "modeling_xtrimopglm.xTrimoPGLMForCasualLM",
    "AutoModelForMaskedLM": "modeling_xtrimopglm.xTrimoPGLMForMaskedLM",
    "AutoModelForSequenceClassification": "modeling_xtrimopglm.xTrimoPGLMForSequenceClassification",
    "AutoModelForTokenClassification": "modeling_xtrimopglm.xTrimoPGLMForTokenClassification"
  },
  "bias_dropout_fusion": true,
  "deepnorm": true,
  "experts_per_token": 0,
  "ffn_hidden_size": 6832,
  "fp32_residual_connection": false,
  "glu_activation": "geglu",
  "head_num": 1,
  "hidden_dropout": 0.0,
  "hidden_size": 2560,
  "initializer_range": 0.02,
  "is_causal": true,
  "kv_channels": 64,
  "layernorm_epsilon": 1e-05,
  "model_type": "xTrimoPGLM",
  "moe": false,
  "multi_query_attention": false,
  "multi_query_group_num": 1,
  "num_attention_heads": 40,
  "num_experts": 0,
  "num_layers": 36,
  "padded_vocab_size": 128,
  "post_layer_norm": true,
  "quantization_bit": 0,
  "rmsnorm": false,
  "rotary_embedding_2d": false,
  "seq_length": 1024,
  "torch_dtype": "float32",
  "transformers_version": "4.41.2",
  "untie_head": false,
  "use_cache": true,
  "use_pytorch_sdpa": true,
  "vocab_size": 128
}