large files

.gitattributes

@@ -32,3 +32,10 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+checkpoints/ filter=lfs diff=lfs merge=lfs -text
+checkpoints/freematch_tune.index filter=lfs diff=lfs merge=lfs -text
+checkpoints/mixmatch.data-00000-of-00001 filter=lfs diff=lfs merge=lfs -text
+checkpoints/mixmatch.index filter=lfs diff=lfs merge=lfs -text
+checkpoints/fixmatch_tune.data-00000-of-00001 filter=lfs diff=lfs merge=lfs -text
+checkpoints/fixmatch_tune.index filter=lfs diff=lfs merge=lfs -text
+checkpoints/freematch_tune.data-00000-of-00001 filter=lfs diff=lfs merge=lfs -text

app.py

@@ -33,9 +33,16 @@ def ssl_predict(in_text, model_type):
         truncation=True,
         return_tensors="tf"
     )
-    if model_type == "freematch":
+
+    if model_type == "fixmatch":
+        model = FixMatchTune(encoder_name="readerbench/RoBERT-base")
+        model.load_weights("./checkpoints/fixmatch_tune")
+    elif model_type == "freematch":
         model = FixMatchTune(encoder_name="andrei-saceleanu/ro-offense-freematch")
         model.cls_head.load_weights("./checkpoints/freematch_tune")
+    elif model_type == "mixmatch":
+        model = MixMatch(encoder_name="andrei-saceleanu/ro-offense-mixmatch")
+        model.cls_head.load_weights("./checkpoints/mixmatch")
 
     preds, _ = model([toks["input_ids"],toks["attention_mask"]], training=False)
     probs = list(preds[0].numpy())

checkpoints/fixmatch_tune.data-00000-of-00001

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8dfdc2f8ad0f036e0bfad3676782816272e98fdf63f09d54883a768084451f8
+size 461147136

checkpoints/fixmatch_tune.index

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:208a951860ca39c7b24278e111864cfca1ec65bcda0ab9ba90f9fa4e052341a2
+size 14764

checkpoints/mixmatch.data-00000-of-00001

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5520588c6d43a9fda3dd5b111f92366fd4d855f321393b01de5d014f4bbe76f1
+size 855091

checkpoints/mixmatch.index

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:699017f5ef5ef36f5e8dad59fa659290960bc36aca0936ca340492add17b23e3
+size 518

models.py

@@ -1,8 +1,10 @@
+"""Model definitions"""
 import tensorflow as tf
 from transformers import TFAutoModel
 
 
 class FixMatchTune(tf.keras.Model):
+    """fixmatch"""
     def __init__(
         self,
         encoder_name="readerbench/RoBERT-base",
@@ -35,3 +37,28 @@ class FixMatchTune(tf.keras.Model):
         weak_preds = self.cls_head(weaks,training=training)
 
         return weak_preds, strong_preds
+
+class MixMatch(tf.keras.Model):
+    """mixmatch"""
+    def __init__(self,bert_model="readerbench/RoBERT-base",num_classes=4,**kwargs):
+        super(MixMatch,self).__init__(**kwargs)
+        self.bert = TFAutoModel.from_pretrained(bert_model)
+
+        self.num_classes = num_classes
+
+        self.cls_head = tf.keras.Sequential([
+            tf.keras.layers.Dense(256,activation="relu"),
+            tf.keras.layers.Dropout(0.2),
+            tf.keras.layers.Dense(64,activation="relu"),
+            tf.keras.layers.Dense(self.num_classes, activation="softmax")
+        ])
+
+        self.augment = tf.keras.layers.GaussianNoise(stddev=2)
+
+    def call(self, inputs, training):
+        ids, mask = inputs
+
+        embeds = self.bert(input_ids=ids, attention_mask=mask,training=training).pooler_output
+        augs = self.augment(embeds,training=training)
+
+        return self.cls_head(augs,training=training)