hyperparameter tuning step 19519

README.md

@@ -18,7 +18,7 @@ metrics:
 
 ## Model description
 
-[sahajBERT](https://huggingface.co/neuropark/sahajBERT-NER) fine-tuned for NER using the bengali of [WikiANN ](https://huggingface.co/datasets/wikiann). 
+[sahajBERT](https://huggingface.co/neuropark/sahajBERT-NER) fine-tuned for NER using the bengali split of [WikiANN ](https://huggingface.co/datasets/wikiann). 
 
 Named Entities predicted by the model:
 
@@ -60,7 +60,7 @@ WIP
 
 ## Training data
 
-The model was initialized it with pre-trained weights of [sahajBERT](https://huggingface.co/neuropark/sahajBERT-NER) at step TODO_REPLACE_BY_STEP_NAME and trained on the bengali of [WikiANN ](https://huggingface.co/datasets/wikiann)
+The model was initialized it with pre-trained weights of [sahajBERT](https://huggingface.co/neuropark/sahajBERT-NER) at step 19519 and trained on the bengali of [WikiANN ](https://huggingface.co/datasets/wikiann)
 
 ## Training procedure
 
@@ -73,16 +73,12 @@ Coming soon!
 
 ## Eval results
 
-accuracy: 0.9756540697674418
-
-f1: 0.9570102589154861
-
-loss: 0.13705264031887054
-
-precision: 0.9518950437317785
-
-recall: 0.962180746561886
 
+loss: 0.11714419722557068
+accuracy: 0.9772286821705426
+precision: 0.9585365853658536
+recall: 0.9651277013752456
+f1 : 0.9618208516886931
 
 
 ### BibTeX entry and citation info

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "albertvillanova/autonlp-wikiann-entity_extraction-1e67664-1301123",
+  "_name_or_path": "albertvillanova/autonlp-baselines-wikiann-entity_extraction-1341171",
   "_num_labels": 7,
   "architectures": [
     "AlbertForTokenClassification"
@@ -36,7 +36,7 @@
     "6": 6
   },
   "layer_norm_eps": 1e-12,
-  "max_length": 96,
+  "max_length": 128,
   "max_position_embeddings": 512,
   "model_type": "albert",
   "net_structure_type": 0,

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:93fef3df43e9670b0c26ae9f82a57a1b7c7e9336c8553c9ed5918b97a413b264
+oid sha256:042f8d9caceb37cadfe054e9e78c8088f7ec98f8c62eb2db5238bf3a71de30f8
 size 67605209