Upload folder using huggingface_hub

config.json

@@ -1,108 +1,44 @@
 {
-  "allow_embedding_resizing": true,
+  "_name_or_path": "RexBERT-mini",
   "architectures": [
     "ModernBertForMaskedLM"
   ],
   "attention_bias": false,
   "attention_dropout": 0.0,
-  "attention_layer": "rope",
-  "attention_probs_dropout_prob": 0.0,
-  "attn_out_bias": false,
-  "attn_out_dropout_prob": 0.1,
-  "attn_qkv_bias": false,
-  "bert_layer": "prenorm",
   "bos_token_id": 50281,
-  "causal_mask": false,
   "classifier_activation": "gelu",
   "classifier_bias": false,
-  "classifier_dropout": null,
-  "classifier_pooling": "cls",
+  "classifier_dropout": 0.0,
+  "classifier_pooling": "mean",
   "cls_token_id": 50281,
-  "compile_model": true,
   "decoder_bias": true,
-  "deterministic_fa2": false,
   "deterministic_flash_attn": false,
-  "embed_dropout_prob": 0.0,
-  "embed_norm": true,
   "embedding_dropout": 0.0,
-  "embedding_layer": "sans_pos",
-  "encoder_layer": "base",
   "eos_token_id": 50282,
-  "final_norm": true,
   "global_attn_every_n_layers": 3,
   "global_rope_theta": 160000.0,
   "gradient_checkpointing": false,
-  "head_class_act": "silu",
-  "head_class_bias": false,
-  "head_class_dropout": 0.0,
-  "head_class_norm": false,
-  "head_pred_act": "gelu",
-  "head_pred_bias": false,
-  "head_pred_dropout": 0.0,
-  "head_pred_norm": true,
-  "hidden_act": "gelu",
   "hidden_activation": "gelu",
-  "hidden_dropout_prob": 0.1,
   "hidden_size": 512,
-  "init_cutoff_factor": 2.0,
-  "init_method": "full_megatron",
-  "init_small_embedding": false,
-  "init_std": 0.02,
-  "initial_attention_layer": null,
-  "initial_bert_layer": null,
-  "initial_mlp_layer": null,
   "initializer_cutoff_factor": 2.0,
   "initializer_range": 0.02,
   "intermediate_size": 768,
-  "layer_norm_eps": 1e-12,
+  "layer_norm_eps": 1e-05,
   "local_attention": 128,
-  "local_attn_rotary_emb_base": 160000.0,
-  "local_attn_rotary_emb_dim": null,
-  "local_rope_theta": 10000.0,
-  "loss_function": "fa_cross_entropy",
-  "loss_kwargs": {
-    "reduction": "mean"
-  },
-  "masked_prediction": true,
-  "max_position_embeddings": 512,
+  "local_rope_theta": 160000.0,
+  "max_position_embeddings": 7999,
   "mlp_bias": false,
   "mlp_dropout": 0.0,
-  "mlp_dropout_prob": 0.0,
-  "mlp_in_bias": false,
-  "mlp_layer": "glu",
-  "mlp_out_bias": false,
   "model_type": "modernbert",
   "norm_bias": false,
   "norm_eps": 1e-05,
-  "norm_kwargs": {
-    "bias": false,
-    "eps": 1e-05
-  },
-  "normalization": "layernorm",
   "num_attention_heads": 8,
   "num_hidden_layers": 19,
-  "num_initial_layers": 1,
-  "pad_logits": false,
-  "pad_token_id": 0,
-  "padding": "unpadded",
-  "pooling_type": "cls",
-  "position_embedding_type": "absolute",
-  "repad_logits_with_grad": false,
-  "rotary_emb_base": 160000.0,
-  "rotary_emb_dim": 64,
-  "rotary_emb_interleaved": false,
-  "rotary_emb_scale_base": null,
+  "pad_token_id": 50283,
+  "position_embedding_type": "sans_pos",
   "sep_token_id": 50282,
-  "skip_first_prenorm": true,
-  "sliding_window": 128,
-  "sparse_pred_ignore_index": -100,
-  "sparse_prediction": false,
+  "tie_word_embeddings": true,
   "torch_dtype": "float32",
-  "transformers_version": "4.55.1",
-  "type_vocab_size": 2,
-  "unpad_embeddings": true,
-  "use_cache": true,
-  "use_fa2": true,
-  "use_sdpa_attn_mask": false,
+  "transformers_version": "4.48.0",
   "vocab_size": 50368
-}
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b4f41f585604367d131be65ed261398d1d5bfa3f3169451ebc349148fe0e8e45
+size 273868543

special_tokens_map.json

@@ -34,4 +34,4 @@
     "rstrip": false,
     "single_word": false
   }
-}
+}

tokenizer_config.json

@@ -931,15 +931,14 @@
   },
   "clean_up_tokenization_spaces": true,
   "cls_token": "[CLS]",
-  "extra_special_tokens": {},
   "mask_token": "[MASK]",
   "model_input_names": [
     "input_ids",
     "attention_mask"
   ],
-  "model_max_length": 8192,
+  "model_max_length": 7999,
   "pad_token": "[PAD]",
   "sep_token": "[SEP]",
   "tokenizer_class": "PreTrainedTokenizerFast",
   "unk_token": "[UNK]"
-}
+}