Training in progress, epoch 1

config.json

@@ -1,6 +1,6 @@
 {
   "architectures": [
-    "RobertaForMaskedLM"
+    "FlashRobertaForMaskedLM"
   ],
   "attention_probs_dropout_prob": 0.1,
   "bos_token_id": 0,
@@ -19,8 +19,8 @@
   "pad_token_id": 1,
   "pooler_hidden_act": "silu",
   "position_embedding_type": "absolute",
-  "torch_dtype": "float32",
-  "transformers_version": "4.41.2",
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.42.0.dev0",
   "type_vocab_size": 2,
   "use_cache": true,
   "vocab_size": 3072

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3a807a85090087696c326eba8b645196db37612c3f7be1e5e388fbc8b52f34a5
-size 353645096
+oid sha256:9876ceebf51c41ca122499b3542d532e6d6744e2714a2c26ced3c16b735f51ee
+size 178016056

runs/Jun25_02-30-59_hoang/events.out.tfevents.1719282704.hoang

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a8f0678f011ab89f79ac3a2dbd2a49e55935e76574dacc569b0d0c8195bac8f9
+size 5600

runs/Jun25_02-37-41_hoang/events.out.tfevents.1719283108.hoang

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0bd378e575d257a82caa8ca2557a75344e06f5d42c2af035c5438f8bd63c4205
+size 120014

runs/Jun25_04-40-01_hoang/events.out.tfevents.1719290440.hoang

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e3a7081d2645ff4881745c0d677b4471659ec31ca0dc9ef53a8518e2352d99ee
+size 4767

runs/Jun25_04-46-37_hoang/events.out.tfevents.1719290844.hoang

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6617a5a17e46ead435060186e3fe000dc6daa521a2f6209cf13149f5df797be2
+size 120008

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5838f3493d6846611c66c0714e817a52a8ffb7e5665820e197951cb458ff2ef9
+oid sha256:7112ed3718ef435e6b11337c36d5229d634449a4d982eabb28f5fa88ac6bc952
 size 6328