Upload ModernALBERTForMaskedLM

config.json

@@ -8,26 +8,26 @@
   "dtype": "float32",
   "embedding_size": 128,
   "expert_intermediate_size": 2624,
-  "group_depth": 4,
+  "group_depth": 3,
   "hidden_act": "gelu",
   "hidden_dropout_prob": 0.1,
   "hidden_size": 768,
   "initializer_range": 0.02,
   "intermediate_size": 1152,
   "layer_norm_eps": 1e-06,
-  "load_balancing_loss_coef": 0.01,
+  "load_balancing_loss_coef": 0.2,
   "lora_alpha": 32,
   "lora_rank": 16,
   "max_position_embeddings": 8192,
   "model_type": "ModernALBERT",
   "num_attention_heads": 16,
-  "num_expert_modules": 2,
+  "num_expert_modules": 3,
   "num_experts": 8,
   "num_hidden_layers": 12,
   "pad_token_id": 0,
   "router_jitter_noise": 0.01,
   "top_k": 2,
-  "transformers_version": "4.56.0",
+  "transformers_version": "4.57.0",
   "use_adapter": true,
   "use_cache": true,
   "use_moa": true,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b3343ee1d4254212ad88ff2b5916776de54192e91ddc15fa0d195bd4b1f69893
-size 128468080
+oid sha256:221e95dc35981f9dcfb34a2cf55a0d6f0b2d8f2424ddffb8871cdd033af9c521
+size 168188440