commit files to HF hub

config.json

@@ -0,0 +1,66 @@
+{
+  "_name_or_path": "tgieruc/Heritage-in-Digital-Age-distilbert-base-uncased-expression-rating",
+  "activation": "gelu",
+  "architectures": [
+    "DistilBertForSequenceClassification"
+  ],
+  "attention_dropout": 0.1,
+  "custom_pipelines": {
+    "expression-ranking": {
+      "default": {
+        "model": {
+          "pt": [
+            "tgieruc/Heritage-in-Digital-Age-distilbert-base-uncased-expression-rating",
+            "expression rating"
+          ]
+        }
+      },
+      "impl": "save_pipeline.ExpressionRankingPipeline",
+      "pt": [
+        "AutoModelForSequenceClassification"
+      ],
+      "tf": []
+    }
+  },
+  "dim": 768,
+  "dropout": 0.1,
+  "hidden_dim": 3072,
+  "id2label": {
+    "0": 0,
+    "1": 1,
+    "2": 2,
+    "3": 3,
+    "4": 4,
+    "5": 5,
+    "6": 6,
+    "7": 7,
+    "8": 8,
+    "9": 9
+  },
+  "initializer_range": 0.02,
+  "label2id": {
+    "0": 0,
+    "1": 1,
+    "2": 2,
+    "3": 3,
+    "4": 4,
+    "5": 5,
+    "6": 6,
+    "7": 7,
+    "8": 8,
+    "9": 9
+  },
+  "max_position_embeddings": 512,
+  "model_type": "distilbert",
+  "n_heads": 12,
+  "n_layers": 6,
+  "pad_token_id": 0,
+  "problem_type": "single_label_classification",
+  "qa_dropout": 0.1,
+  "seq_classif_dropout": 0.2,
+  "sinusoidal_pos_embds": false,
+  "tie_weights_": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.22.0",
+  "vocab_size": 30522
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3277c0b553d4698d717c1b7a66cca0d55a84ec55af78b7e5ec0bc3ad06ba729b
+size 267877489

save_pipeline.py

@@ -0,0 +1,68 @@
+from transformers import Pipeline
+from transformers import AutoTokenizer
+    
+import numpy as np
+
+
+def softmax(outputs):
+    maxes = np.max(outputs, axis=-1, keepdims=True)
+    shifted_exp = np.exp(outputs - maxes)
+    return shifted_exp / shifted_exp.sum(axis=-1, keepdims=True)
+
+
+
+class HeritageDigitalAgePipeline(Pipeline):
+    def _sanitize_parameters(self, **kwargs):
+        preprocess_kwargs = {}
+        if "expression1" in kwargs:
+            preprocess_kwargs["caption"] = str(kwargs["caption"]).lower()
+        if "expression2" in kwargs:
+            preprocess_kwargs["title"] = str(kwargs["title"]).lower()
+        return preprocess_kwargs, {}, {}
+
+    def preprocess(self, inputs, maybe_arg=2):
+        sep_token = "[SEP]"
+        tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased", use_fast=True)
+        model_input = tokenizer.encode(inputs['caption'] + sep_token + inputs['title'], return_tensors='pt', add_special_tokens=True,  truncation=True)
+        return {"model_input": model_input}
+
+    def _forward(self, model_inputs):
+        # model_inputs == {"model_input": model_input}
+        return self.model(model_inputs['model_input'])
+        
+    def postprocess(self, model_outputs):
+        logits = model_outputs.logits[0].numpy()
+        probabilities = softmax(logits)
+
+        best_class = np.argmax(probabilities)
+        label = self.model.config.id2label[best_class]
+        score = probabilities[best_class].item()
+        logits = logits.tolist()
+
+        return {"label": label, "score": score, "logits": logits}
+
+
+class ExpressionRankingPipeline(Pipeline):
+    def _sanitize_parameters(self, **kwargs):
+        preprocess_kwargs = {}
+        return preprocess_kwargs, {}, {}
+
+    def preprocess(self, inputs, maybe_arg=2):
+        tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased", use_fast=True)
+        model_input = tokenizer(inputs, truncation=True, padding="max_length", max_length=256, return_tensors="pt")
+
+        return {"model_input": model_input}
+
+    def _forward(self, model_inputs):
+        return self.model(**model_inputs['model_input'])
+        
+    def postprocess(self, model_outputs):
+        logits = model_outputs.logits[0].numpy()
+        probabilities = softmax(logits)
+
+        best_class = np.argmax(probabilities)
+        label = self.model.config.id2label[best_class]
+        score = probabilities[best_class].item()
+        logits = logits.tolist()
+
+        return {"label": label, "score": score, "logits": logits}

special_tokens_map.json

@@ -0,0 +1,7 @@
+{
+  "cls_token": "[CLS]",
+  "mask_token": "[MASK]",
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "unk_token": "[UNK]"
+}

tokenizer_config.json

@@ -0,0 +1,14 @@
+{
+  "cls_token": "[CLS]",
+  "do_lower_case": true,
+  "mask_token": "[MASK]",
+  "model_max_length": 512,
+  "name_or_path": "tgieruc/Heritage-in-Digital-Age-distilbert-base-uncased-expression-rating",
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "special_tokens_map_file": null,
+  "strip_accents": null,
+  "tokenize_chinese_chars": true,
+  "tokenizer_class": "DistilBertTokenizer",
+  "unk_token": "[UNK]"
+}