{
  "architectures": [
    "PragyaForCausalLM"
  ],
  "auto_map": {
    "AutoConfig": "configuration_pragya.PragyaConfig",
    "AutoModelForCausalLM": "modeling_pragya.PragyaForCausalLM"
  },
  "bias": false,
  "block_size": 1024,
  "d_rope": null,
  "dropout": 0.0,
  "engram_compressed_vocab": 9807,
  "engram_embed_per_ngram": 128,
  "engram_kernel_size": 4,
  "engram_layer_ids": [
    1,
    8
  ],
  "engram_max_ngram": 3,
  "engram_n_heads": 4,
  "engram_seed": 0,
  "engram_vocab_size": [
    4096,
    4096
  ],
  "ffn_mult": 2,
  "gen_headroom": 128,
  "gpt_top_k": 64,
  "gradient_checkpointing": false,
  "int8_group_size": 128,
  "max_gen_tokens": 1024,
  "mod_budget_coeff": 0.05,
  "mod_capacity": 0.25,
  "mod_gate_entropy_coeff": 0.0,
  "mod_hard_eval": false,
  "model_type": "pragya_e2",
  "moe_fixed_capacity": null,
  "moe_top_k": 2,
  "mtp_depth": 1,
  "mtp_lambda": 0.1,
  "n_embd": 384,
  "n_experts": 8,
  "n_head": 12,
  "n_layer": 16,
  "n_shared": 1,
  "num_kv_heads": 6,
  "pattern": "dense",
  "sliding_window_size": 512,
  "tp_size": 1,
  "transformers_version": "5.20.1",
  "use_engram": true,
  "use_int8": false,
  "use_moe": false,
  "use_mtp": false,
  "use_nvfp4": false,
  "vocab_size": 12288
}