Upload config.json

config.json

@@ -1,10 +1,15 @@
 {
-  "_name_or_path": "google/gemma-7b",
+  "_name_or_path": "Crystalcareai/GemMoE-Beta-1",
   "architectures": [
-    "MixtralForCausalLM"
+    "GemmoeForCausalLM"
   ],
   "attention_bias": false,
   "attention_dropout": 0.0,
+  "auto_map": {
+    "AutoConfig": "Crystalcareai/GemMoE-Beta-1--configuration_gemmoe.GemmoeConfig",
+    "AutoModelForCausalLM": "Crystalcareai/GemMoE-Beta-1--modeling_gemmoe.GemmoeForCausalLM",
+    "AutoTokenizer": "Crystalcareai/GemMoE-Beta-1--tokenization_gemmoe.GemmoeTokenizer",
+  },
   "bos_token_id": 2,
   "eos_token_id": 1,
   "head_dim": 256,
@@ -13,7 +18,7 @@
   "initializer_range": 0.02,
   "intermediate_size": 24576,
   "max_position_embeddings": 8192,
-  "model_type": "mixtral",
+  "model_type": "gemmoe",
   "num_attention_heads": 16,
   "num_experts_per_tok": 2,
   "num_hidden_layers": 28,
@@ -24,11 +29,10 @@
   "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": true,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.38.2",
-  "use_cache": true,
+  "use_cache": false,
   "vocab_size": 256000
 }