Update config.json

config.json

@@ -1,3 +1,4 @@
+
 {
   "_name_or_path": "damerajee/hathi-moe-test",
   "architectures": [
@@ -7,32 +8,31 @@
   "attention_dropout": 0.0,
   "auto_map": {
     "AutoConfig": "damerajee/hathi-moe-test--configuration_Hixtral.HixtralConfig",
-    "AutoModelForCausalLM": "damerajee/hathi-moe-test--modeling_Hixtral.HixtralForCausalLM",
-
+    "AutoModelForCausalLM": "damerajee/hathi-moe-test--modeling_Hixtral.HixtralForCausalLM"
   },
-  "bos_token_id": 1,
-  "eos_token_id": 2,
-  "hidden_act": "silu",
-  "hidden_size": 4096,
+  "bos_token_id": 2,
+  "eos_token_id": 1,
+  "head_dim": 256,
+  "hidden_act": "gelu",
+  "hidden_size": 3072,
   "initializer_range": 0.02,
-  "intermediate_size": 11008,
-  "max_position_embeddings": 4096,
-  "model_type": "hixtral",
-  "num_attention_heads": 32,
+  "intermediate_size": 24576,
+  "max_position_embeddings": 8192,
+  "model_type": "gemmoe",
+  "num_attention_heads": 16,
   "num_experts_per_tok": 2,
-  "num_hidden_layers": 32,
-  "num_key_value_heads": 32,
-  "num_local_experts": 4,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 16,
+  "num_local_experts": 8,
   "output_router_logits": false,
-  "pretraining_tp": 1,
-  "rms_norm_eps": 1e-05,
+  "pad_token_id": 0,
+  "rms_norm_eps": 1e-06,
   "rope_scaling": null,
   "rope_theta": 10000.0,
-  "router_aux_loss_coef": 0.001,
+  "router_aux_loss_coef": 0.02,
   "sliding_window": null,
-  "tie_word_embeddings": false,
-  "torch_dtype": "float16",
+  "torch_dtype": "bfloat16",
   "transformers_version": "4.38.2",
-  "use_cache": true,
+  "use_cache": false,
   "vocab_size": 48064
-}
+}