Training and fine-tuning embeddings complete!

Runs without errors.
Next must generate two random words for analogies.

ag_news_model/1_Pooling/config.json

@@ -0,0 +1,7 @@
+{
+  "word_embedding_dimension": 384,
+  "pooling_mode_cls_token": false,
+  "pooling_mode_mean_tokens": true,
+  "pooling_mode_max_tokens": false,
+  "pooling_mode_mean_sqrt_len_tokens": false
+}

ag_news_model/README.md

@@ -0,0 +1,88 @@
+---
+pipeline_tag: sentence-similarity
+tags:
+- sentence-transformers
+- feature-extraction
+- sentence-similarity
+
+---
+
+# {MODEL_NAME}
+
+This is a [sentence-transformers](https://www.SBERT.net) model: It maps sentences & paragraphs to a 384 dimensional dense vector space and can be used for tasks like clustering or semantic search.
+
+<!--- Describe your model here -->
+
+## Usage (Sentence-Transformers)
+
+Using this model becomes easy when you have [sentence-transformers](https://www.SBERT.net) installed:
+
+```
+pip install -U sentence-transformers
+```
+
+Then you can use the model like this:
+
+```python
+from sentence_transformers import SentenceTransformer
+sentences = ["This is an example sentence", "Each sentence is converted"]
+
+model = SentenceTransformer('{MODEL_NAME}')
+embeddings = model.encode(sentences)
+print(embeddings)
+```
+
+
+
+## Evaluation Results
+
+<!--- Describe how your model was evaluated -->
+
+For an automated evaluation of this model, see the *Sentence Embeddings Benchmark*: [https://seb.sbert.net](https://seb.sbert.net?model_name={MODEL_NAME})
+
+
+## Training
+The model was trained with the parameters:
+
+**DataLoader**:
+
+`torch.utils.data.dataloader.DataLoader` of length 2400 with parameters:
+```
+{'batch_size': 25, 'sampler': 'torch.utils.data.sampler.RandomSampler', 'batch_sampler': 'torch.utils.data.sampler.BatchSampler'}
+```
+
+**Loss**:
+
+`sentence_transformers.losses.BatchHardSoftMarginTripletLoss.BatchHardSoftMarginTripletLoss` 
+
+Parameters of the fit()-Method:
+```
+{
+    "epochs": 10,
+    "evaluation_steps": 0,
+    "evaluator": "NoneType",
+    "max_grad_norm": 1,
+    "optimizer_class": "<class 'torch.optim.adamw.AdamW'>",
+    "optimizer_params": {
+        "lr": 2e-05
+    },
+    "scheduler": "WarmupLinear",
+    "steps_per_epoch": null,
+    "warmup_steps": 10000,
+    "weight_decay": 0.01
+}
+```
+
+
+## Full Model Architecture
+```
+SentenceTransformer(
+  (0): Transformer({'max_seq_length': 256, 'do_lower_case': False}) with Transformer model: BertModel 
+  (1): Pooling({'word_embedding_dimension': 384, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False})
+  (2): Normalize()
+)
+```
+
+## Citing & Authors
+
+<!--- Describe where people can find more information -->

ag_news_model/config.json

@@ -0,0 +1,26 @@
+{
+  "_name_or_path": "/home/smhavens/.cache/torch/sentence_transformers/sentence-transformers_all-MiniLM-L6-v2/",
+  "architectures": [
+    "BertModel"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "classifier_dropout": null,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 384,
+  "initializer_range": 0.02,
+  "intermediate_size": 1536,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "bert",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 6,
+  "pad_token_id": 0,
+  "position_embedding_type": "absolute",
+  "torch_dtype": "float32",
+  "transformers_version": "4.35.2",
+  "type_vocab_size": 2,
+  "use_cache": true,
+  "vocab_size": 30522
+}

ag_news_model/config_sentence_transformers.json

@@ -0,0 +1,7 @@
+{
+  "__version__": {
+    "sentence_transformers": "2.0.0",
+    "transformers": "4.6.1",
+    "pytorch": "1.8.1"
+  }
+}

ag_news_model/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1167225afc7fa8d939844ba7a4715244880742fc1c279ef49127fe2dd36a796
+size 90864192

ag_news_model/modules.json

@@ -0,0 +1,20 @@
+[
+  {
+    "idx": 0,
+    "name": "0",
+    "path": "",
+    "type": "sentence_transformers.models.Transformer"
+  },
+  {
+    "idx": 1,
+    "name": "1",
+    "path": "1_Pooling",
+    "type": "sentence_transformers.models.Pooling"
+  },
+  {
+    "idx": 2,
+    "name": "2",
+    "path": "2_Normalize",
+    "type": "sentence_transformers.models.Normalize"
+  }
+]

ag_news_model/sentence_bert_config.json

@@ -0,0 +1,4 @@
+{
+  "max_seq_length": 256,
+  "do_lower_case": false
+}

ag_news_model/special_tokens_map.json

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

ag_news_model/tokenizer_config.json

@@ -0,0 +1,64 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "[PAD]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100": {
+      "content": "[UNK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "101": {
+      "content": "[CLS]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "102": {
+      "content": "[SEP]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "103": {
+      "content": "[MASK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "clean_up_tokenization_spaces": true,
+  "cls_token": "[CLS]",
+  "do_basic_tokenize": true,
+  "do_lower_case": true,
+  "mask_token": "[MASK]",
+  "max_length": 128,
+  "model_max_length": 512,
+  "never_split": null,
+  "pad_to_multiple_of": null,
+  "pad_token": "[PAD]",
+  "pad_token_type_id": 0,
+  "padding_side": "right",
+  "sep_token": "[SEP]",
+  "stride": 0,
+  "strip_accents": null,
+  "tokenize_chinese_chars": true,
+  "tokenizer_class": "BertTokenizer",
+  "truncation_side": "right",
+  "truncation_strategy": "longest_first",
+  "unk_token": "[UNK]"
+}

app.py

@@ -19,7 +19,7 @@ import sys
 
 # !pip install https://huggingface.co/spacy/en_core_web_sm/resolve/main/en_core_web_sm-any-py3-none-any.whl
 # subprocess.check_call([sys.executable, '-m', 'pip', 'install', 'https://huggingface.co/spacy/en_core_web_sm/resolve/main/en_core_web_sm-any-py3-none-any.whl'])
-tokenizer = AutoTokenizer.from_pretrained("bert-base-cased")
+# tokenizer = AutoTokenizer.from_pretrained('sentence-transformers/all-MiniLM-L6-v2')
 nltk.download('stopwords')
 nlp = spacy.load("en_core_web_sm")
 stops = stopwords.words("english")
@@ -31,7 +31,7 @@ answer = "Pizza"
 
 #Mean Pooling - Take attention mask into account for correct averaging
 def mean_pooling(model_output, attention_mask):
-    token_embeddings = model_output[0] #First element of model_output contains all token embeddings
+    token_embeddings = model_output['token_embeddings'] #First element of model_output contains all token embeddings
     input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
     return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)
 
@@ -49,8 +49,8 @@ def normalize(comment, lowercase, remove_stopwords):
     return " ".join(lemmatized)
 
 
-def tokenize_function(examples):
-    return tokenizer(examples["text"])
+# def tokenize_function(examples):
+#     return tokenizer(examples["text"])
 
 
 def compute_metrics(eval_pred):
@@ -181,28 +181,56 @@ def finetune(train_dataloader):
     
     # trainer.train()
     
+def embeddings():
+    model = SentenceTransformer("ag_news_model")
+    device = torch.device('cuda:0')
+    model = model.to(device)
     sentences = ["This is an example sentence", "Each sentence is converted"]
 
     # model = SentenceTransformer('sentence-transformers/all-MiniLM-L6-v2')
     embeddings = model.encode(sentences)
-    print(embeddings)
+    # print(embeddings)
     
     # Sentences we want sentence embeddings for
     sentences = ['This is an example sentence', 'Each sentence is converted']
 
     # Load model from HuggingFace Hub
-    # tokenizer = AutoTokenizer.from_pretrained('sentence-transformers/all-MiniLM-L6-v2')
+    tokenizer = AutoTokenizer.from_pretrained('ag_news_model')
     # model = AutoModel.from_pretrained('sentence-transformers/all-MiniLM-L6-v2')
 
     # Tokenize sentences
     encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')
 
+    # print(model.device)
+    # print(encoded_input["input_ids"].device)
+    # print(encoded_input["attention_mask"].device)
+    # print(encoded_input["token_type_ids"].device)
+    encoded_input["input_ids"] = encoded_input["input_ids"].to(device)
+    encoded_input["attention_mask"] = encoded_input["attention_mask"].to(device)
+    encoded_input['token_type_ids'] = encoded_input['token_type_ids'].to(device)
+    # print(encoded_input)
+    
+    # print(encoded_input["input_ids"].device)
+    # print(encoded_input["attention_mask"].device)
+    # print(encoded_input["token_type_ids"].device)
+    
+    encoded_input['input'] = {'input_ids':encoded_input['input_ids'], 'attention_mask':encoded_input['attention_mask']}
+    
+    #  + encoded_input['token_type_ids'] + encoded_input['attention_mask']
+    del encoded_input['input_ids']
+    del encoded_input['token_type_ids']
+    del encoded_input['attention_mask']
+
+    # print(encoded_input)
+    
+    # encoded_input.to(device)
     # Compute token embeddings
     with torch.no_grad():
         model_output = model(**encoded_input)
 
+    print(model_output)
     # Perform pooling
-    sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])
+    sentence_embeddings = mean_pooling(model_output, encoded_input['input']["attention_mask"])
 
     # Normalize embeddings
     sentence_embeddings = F.normalize(sentence_embeddings, p=2, dim=1)
@@ -238,7 +266,8 @@ def main():
     answer = "Moon"
     global guesses
     
-    num_rows, data_type, value, example, embeddings = training()
+    # num_rows, data_type, value, example, embeddings = training()
+    sent_embeddings = embeddings()
     
     prompt = f"{word1} is to {word2} as {word3} is to ____"
     with gr.Blocks() as iface:
@@ -250,9 +279,7 @@ def main():
         with gr.Accordion("Open for previous guesses"):
             text_guesses = gr.Textbox()
         with gr.Tab("Testing"):
-            gr.Markdown(f"""Number of rows in dataset is {num_rows}, with each having type {data_type} and value {value}.
-                        An example is {example}.
-                        The Embeddings are {embeddings}.""")
+            gr.Markdown(f"""The Embeddings are {sent_embeddings}.""")
         text_button.click(check_answer, inputs=[text_input], outputs=[text_output, text_guesses])
     # iface = gr.Interface(fn=greet, inputs="text", outputs="text")
     iface.launch()

train.py

@@ -16,10 +16,13 @@ import nltk
 from nltk.corpus import stopwords
 import subprocess
 import sys
+from transformers import DataCollatorWithPadding
+
 
 # !pip install https://huggingface.co/spacy/en_core_web_sm/resolve/main/en_core_web_sm-any-py3-none-any.whl
 # subprocess.check_call([sys.executable, '-m', 'pip', 'install', 'https://huggingface.co/spacy/en_core_web_sm/resolve/main/en_core_web_sm-any-py3-none-any.whl'])
-tokenizer = AutoTokenizer.from_pretrained("bert-base-cased")
+# tokenizer = AutoTokenizer.from_pretrained('sentence-transformers/all-MiniLM-L6-v2')
+# data_collator = DataCollatorWithPadding(tokenizer=tokenizer)
 # nltk.download('stopwords')
 # nlp = spacy.load("en_core_web_sm")
 # stops = stopwords.words("english")
@@ -49,8 +52,8 @@ def mean_pooling(model_output, attention_mask):
 #     return " ".join(lemmatized)
 
 
-def tokenize_function(examples):
-    return tokenizer(examples["text"])
+# def tokenize_function(examples):
+#     return tokenizer(examples["text"], truncation=True)
 
 
 def compute_metrics(eval_pred):
@@ -62,6 +65,8 @@ def compute_metrics(eval_pred):
 
 def training():
     dataset_id = "ag_news"
+    
+    print("GETTING DATASET")
     dataset = load_dataset(dataset_id)
     # dataset = dataset["train"]
     # tokenized_datasets = dataset.map(tokenize_function, batched=True)
@@ -77,6 +82,10 @@ def training():
     # dataset.set_format(type="torch", columns=["input_ids", "token_type_ids", "attention_mask", "label"])
     # dataset.format['type']
     
+    # tokenized_news = dataset.map(tokenize_function, batched=True)
+    
+    # model = AutoModelForSequenceClassification.from_pretrained("sentence-transformers/all-MiniLM-L6-v2", num_labels=2)
+    
     # print(dataset)
     
     train_examples = []
@@ -152,6 +161,8 @@ def finetune(train_dataloader):
     # model = AutoModelForSequenceClassification.from_pretrained("bert-base-cased", num_labels=5)
     model_id = "sentence-transformers/all-MiniLM-L6-v2"
     model = SentenceTransformer(model_id)
+    device = torch.device('cuda:0')
+    model = model.to(device)
     
     # training_args = TrainingArguments(output_dir="test_trainer")
     
@@ -181,35 +192,35 @@ def finetune(train_dataloader):
     
     # trainer.train()
     
-    sentences = ["This is an example sentence", "Each sentence is converted"]
+    # sentences = ["This is an example sentence", "Each sentence is converted"]
 
-    # model = SentenceTransformer('sentence-transformers/all-MiniLM-L6-v2')
-    embeddings = model.encode(sentences)
-    print(embeddings)
+    # # model = SentenceTransformer('sentence-transformers/all-MiniLM-L6-v2')
+    # embeddings = model.encode(sentences)
+    # print(embeddings)
     
-    # Sentences we want sentence embeddings for
-    sentences = ['This is an example sentence', 'Each sentence is converted']
+    # # Sentences we want sentence embeddings for
+    # sentences = ['This is an example sentence', 'Each sentence is converted']
 
-    # Load model from HuggingFace Hub
-    # tokenizer = AutoTokenizer.from_pretrained('sentence-transformers/all-MiniLM-L6-v2')
-    # model = AutoModel.from_pretrained('sentence-transformers/all-MiniLM-L6-v2')
+    # # Load model from HuggingFace Hub
+    # # tokenizer = AutoTokenizer.from_pretrained('sentence-transformers/all-MiniLM-L6-v2')
+    # # model = AutoModel.from_pretrained('sentence-transformers/all-MiniLM-L6-v2')
 
-    # Tokenize sentences
-    encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')
+    # # Tokenize sentences
+    # encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')
 
-    # Compute token embeddings
-    with torch.no_grad():
-        model_output = model(**encoded_input)
+    # # Compute token embeddings
+    # with torch.no_grad():
+    #     model_output = model(**encoded_input)
 
-    # Perform pooling
-    sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])
+    # # Perform pooling
+    # sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])
 
-    # Normalize embeddings
-    sentence_embeddings = F.normalize(sentence_embeddings, p=2, dim=1)
+    # # Normalize embeddings
+    # sentence_embeddings = F.normalize(sentence_embeddings, p=2, dim=1)
 
-    print("Sentence embeddings:")
-    print(sentence_embeddings)
-    return sentence_embeddings
+    # print("Sentence embeddings:")
+    # print(sentence_embeddings)
+    return 0
 
  
 
@@ -231,6 +242,7 @@ def check_answer(guess:str):
         return "Try again!", output
 
 def main():
+    print("BEGIN")
     word1 = "Black"
     word2 = "White"
     word3 = "Sun"