---
dataset_info:
  features:
  - name: index
    dtype: int64
  - name: modelId
    dtype: string
  - name: config_model_type
    dtype: string
  - name: config_architectures
    dtype: string
  - name: config_vocab_size
    dtype: string
  - name: config_torch_dtype
    dtype: string
  - name: config_transformers_version
    dtype: string
  - name: config_hidden_size
    dtype: float64
  - name: config_intermediate_size
    dtype: float64
  - name: config_num_hidden_layers
    dtype: string
  - name: config_num_attention_heads
    dtype: string
  - name: config_num_key_value_heads
    dtype: float64
  - name: config_hidden_act
    dtype: string
  - name: config_attention_dropout
    dtype: float64
  - name: config_use_cache
    dtype: string
  - name: config_max_position_embeddings
    dtype: string
  - name: config_rope_theta
    dtype: string
  - name: config_rope_scaling_type
    dtype: string
  - name: config_rms_norm_eps
    dtype: string
  - name: config_initializer_range
    dtype: string
  - name: config_bos_token_id
    dtype: string
  - name: config_eos_token_id
    dtype: string
  - name: config_tie_word_embeddings
    dtype: string
  - name: config_head_dimension
    dtype: string
  - name: config_gqa_ratio
    dtype: string
  - name: config_context_per_layer
    dtype: string
  - name: config_approx_params_billions
    dtype: string
  - name: is_llama_family
    dtype: string
  - name: is_bert_family
    dtype: string
  - name: is_gpt_family
    dtype: string
  - name: is_t5_family
    dtype: string
  - name: is_whisper_family
    dtype: string
  - name: is_deepseek_family
    dtype: string
  - name: is_mistral_family
    dtype: string
  - name: uses_moe
    dtype: string
  - name: uses_gqa
    dtype: string
  - name: uses_rope
    dtype: string
  - name: uses_quantization
    dtype: string
  - name: uses_lora
    dtype: string
  - name: size_category
    dtype: string
  - name: context_category
    dtype: string
  - name: config_attention_bias
    dtype: string
  - name: config_auto_map_autoconfig
    dtype: string
  - name: config_auto_map_automodel
    dtype: string
  - name: config_auto_map_automodelforcausallm
    dtype: string
  - name: config_ep_size
    dtype: string
  - name: config_first_k_dense_replace
    dtype: string
  - name: config_kv_lora_rank
    dtype: string
  - name: config_layer_norm_eps
    dtype: string
  - name: config_moe_enabled
    dtype: string
  - name: config_moe_expert_utilization
    dtype: string
  - name: config_moe_intermediate_size
    dtype: float64
  - name: config_moe_layer_freq
    dtype: string
  - name: config_n_group
    dtype: float64
  - name: config_n_routed_experts
    dtype: float64
  - name: config_n_shared_experts
    dtype: float64
  - name: config_norm_topk_prob
    dtype: string
  - name: config_num_experts_per_tok
    dtype: float64
  - name: config_num_nextn_predict_layers
    dtype: float64
  - name: config_pad_token_id
    dtype: string
  - name: config_q_lora_rank
    dtype: float64
  - name: config_qk_nope_head_dim
    dtype: float64
  - name: config_qk_rope_head_dim
    dtype: float64
  - name: config_quantization_activation_scheme
    dtype: string
  - name: config_quantization_fmt
    dtype: string
  - name: config_quantization_method
    dtype: string
  - name: config_quantization_weight_block_size
    dtype: string
  - name: config_rope_scaling_beta_fast
    dtype: float64
  - name: config_rope_scaling_beta_slow
    dtype: string
  - name: config_rope_scaling_factor
    dtype: float64
  - name: config_rope_scaling_mscale
    dtype: string
  - name: config_rope_scaling_mscale_all_dim
    dtype: string
  - name: config_rope_scaling_original_max_pos
    dtype: string
  - name: config_routed_scaling_factor
    dtype: string
  - name: config_scoring_func
    dtype: string
  - name: config_topk_group
    dtype: string
  - name: config_topk_method
    dtype: string
  - name: config_v_head_dim
    dtype: string
  splits:
  - name: train
    num_bytes: 8500857
    num_examples: 14566
  download_size: 945908
  dataset_size: 8500857
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---