Upload checkpoint at step 2001

README.md

@@ -0,0 +1,40 @@
+---
+language: multilingual
+license: apache-2.0
+tags:
+- bert
+- modernbert
+- flexbert
+- masked-language-modeling
+---
+
+# Bert
+
+This model is a ModernBERT/FlexBERT checkpoint uploaded during training.
+
+## Training Information
+
+- **Step**: 2001
+- **Epoch**: 0
+- **Samples Seen**: 1070555
+
+## Metrics
+
+
+
+## Model Architecture
+
+This model uses the FlexBERT architecture with modern improvements over traditional BERT.
+
+## Usage
+
+```python
+from transformers import AutoModel, AutoTokenizer
+
+model = AutoModel.from_pretrained("QuangDuy/Bert")
+tokenizer = AutoTokenizer.from_pretrained("QuangDuy/Bert")
+```
+
+## Citation
+
+If you use this model, please cite the ModernBERT paper.

config.json

@@ -0,0 +1,85 @@
+{
+  "allow_embedding_resizing": true,
+  "architectures": [
+    "ModernBertForMaskedLM"
+  ],
+  "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": 1,
+  "classifier_dropout": null,
+  "compile_model": true,
+  "decoder_bias": true,
+  "deterministic_fa2": false,
+  "embed_dropout_prob": 0.0,
+  "embed_norm": true,
+  "embedding_layer": "sans_pos",
+  "encoder_layer": "base",
+  "eos_token_id": 2,
+  "final_norm": true,
+  "global_attn_every_n_layers": 3,
+  "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_dropout_prob": 0.0,
+  "hidden_size": 768,
+  "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_range": 0.02,
+  "intermediate_size": 1152,
+  "layer_norm_eps": 1e-06,
+  "local_attn_rotary_emb_base": -1,
+  "local_attn_rotary_emb_dim": null,
+  "loss_function": "fa_cross_entropy",
+  "loss_kwargs": {
+    "reduction": "mean"
+  },
+  "mask_token_id": 38370,
+  "masked_prediction": true,
+  "max_position_embeddings": 1024,
+  "mlp_dropout_prob": 0.0,
+  "mlp_in_bias": false,
+  "mlp_layer": "glu",
+  "mlp_out_bias": false,
+  "model_type": "bert",
+  "norm_kwargs": {
+    "bias": false,
+    "eps": 1e-05
+  },
+  "normalization": "layernorm",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 6,
+  "num_initial_layers": 1,
+  "pad_logits": false,
+  "pad_token_id": 38369,
+  "padding": "unpadded",
+  "pooling_type": "cls",
+  "position_embedding_type": "absolute",
+  "rotary_emb_base": 10000.0,
+  "rotary_emb_dim": 64,
+  "rotary_emb_interleaved": false,
+  "rotary_emb_scale_base": null,
+  "skip_first_prenorm": true,
+  "sliding_window": 128,
+  "transformers_version": "4.44.1",
+  "type_vocab_size": 2,
+  "unpad_embeddings": true,
+  "use_cache": true,
+  "use_fa2": true,
+  "use_sdpa_attn_mask": false,
+  "vocab_size": 38376
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5dbbaeabd57bfda179693b840bba83d66a0a5e5d06f920e0095e8a218198e2a1
+size 240790377