Training in progress, step 500

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "flaubert/flaubert_small_cased",
+  "_name_or_path": "flaubert/flaubert_base_cased",
   "amp": 1,
   "architectures": [
     "FlaubertForSequenceClassification"
@@ -12,7 +12,7 @@
   "causal": false,
   "clip_grad_norm": 5,
   "dropout": 0.1,
-  "emb_dim": 512,
+  "emb_dim": 768,
   "embed_init_std": 0.02209708691207961,
   "encoder_only": true,
   "end_n_top": 5,
@@ -48,8 +48,8 @@
   "langs": [
     "fr"
   ],
-  "layer_norm_eps": 1e-06,
-  "layerdrop": 0.2,
+  "layer_norm_eps": 1e-12,
+  "layerdrop": 0.0,
   "lg_sampling_factor": -1,
   "lgs": "fr",
   "mask_index": 5,
@@ -64,12 +64,12 @@
     ]
   ],
   "model_type": "flaubert",
-  "n_heads": 8,
+  "n_heads": 12,
   "n_langs": 1,
-  "n_layers": 6,
+  "n_layers": 12,
   "pad_index": 2,
   "pad_token_id": 2,
-  "pre_norm": true,
+  "pre_norm": false,
   "problem_type": "single_label_classification",
   "sample_alpha": 0,
   "share_inout_emb": true,

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:11d5bae3568acafd0ad7c201ac76346bc2c84dd51bb7e11b2ae09562e262c27f
-size 217510949
+oid sha256:e292e633ad3ed12bdceb8f380f90d519bb95b6c89ba57214c5e349a93dd462e5
+size 553012965

runs/May15_12-36-32_00a28a6b8d46/1684154413.3199668/events.out.tfevents.1684154413.00a28a6b8d46.15988.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:549359b9f4b4239893315e8d1d01eb5c446f9c97d6d9ce5ec2398e935748311f
+size 5892

runs/May15_12-36-32_00a28a6b8d46/1684154581.5030372/events.out.tfevents.1684154581.00a28a6b8d46.15988.3

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91fc22b7eebef1af1f0f6a70148a4220d30b59f372481dcd2aeafe294112ab47
+size 5892

runs/May15_12-36-32_00a28a6b8d46/events.out.tfevents.1684154413.00a28a6b8d46.15988.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5496de8eba2f223a5033e6eb11e15daaae37d3508400cd4ea696991bf57779b8
+size 5256

runs/May15_12-36-32_00a28a6b8d46/events.out.tfevents.1684154581.00a28a6b8d46.15988.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80fc071483f6dddf7f923cf5cb4d3a380912f789dde1ca57cd7cc834d8b28ea7
+size 5365

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:88edb5f114647dffe9dd4b391d3a834c10d614ef5efb09f224c6b2182bd46562
+oid sha256:6cfa561034f17a42f3883392bfbd5eb29f288266ae58fc3a8f4b70e2e5eb9fd9
 size 3899