Training in progress, epoch 1

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "answerdotai/ModernBERT-base",
+  "_name_or_path": "answerdotai/ModernBERT-large",
   "architectures": [
     "ModernBertForSequenceClassification"
   ],
@@ -19,14 +19,14 @@
   "global_rope_theta": 160000.0,
   "gradient_checkpointing": false,
   "hidden_activation": "gelu",
-  "hidden_size": 768,
+  "hidden_size": 1024,
   "id2label": {
     "0": "small_llm",
     "1": "large_llm"
   },
   "initializer_cutoff_factor": 2.0,
   "initializer_range": 0.02,
-  "intermediate_size": 1152,
+  "intermediate_size": 2624,
   "label2id": {
     "large_llm": "1",
     "small_llm": "0"
@@ -40,8 +40,8 @@
   "model_type": "modernbert",
   "norm_bias": false,
   "norm_eps": 1e-05,
-  "num_attention_heads": 12,
-  "num_hidden_layers": 22,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
   "pad_token_id": 50283,
   "position_embedding_type": "absolute",
   "problem_type": "single_label_classification",

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:877175f6eee04271d158c805ba3d235d23c297562fc418110eb35c95319724e6
-size 598439784
+oid sha256:b86aa12e93e4c7d6f071505e5367cd7bb3ae85f115393a7e9240542685625da7
+size 1583351632

runs/Feb05_10-54-32_ubuntu-t4/events.out.tfevents.1738752873.ubuntu-t4.901182.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff5bf65c375b2c3ab72132942ecd62eb04c4931453296ec54d0d3cf2b0969df4
+size 5806

runs/Feb05_10-55-06_ubuntu-t4/events.out.tfevents.1738752907.ubuntu-t4.901366.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5c25afe46022c42e462611f02b34a46adcf8f866c2718c60420440d7b5aad06b
+size 5806

runs/Feb05_11-04-41_ubuntu-t4/events.out.tfevents.1738753482.ubuntu-t4.903191.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2de5ab3ccd28c7bffa611f9551fae6897749bb5379d74f7eb709d427b44363db
+size 14137

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8b60d36f1d3533ba6d92660242ad1c9da9eaa1dfe1d37741d516acd57def3bfa
+oid sha256:f64381a525fb162fe9d42e22e1b2eb5bac96df22d92d4db1c31576121f5b2438
 size 5368