Commit
β’
fbadaa0
1
Parent(s):
1c1bd81
Update README.md
Browse files
README.md
CHANGED
@@ -1,77 +1,87 @@
|
|
1 |
---
|
|
|
|
|
|
|
|
|
|
|
2 |
library_name: transformers
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4 |
---
|
5 |
|
6 |
-
# Model Card for
|
7 |
|
8 |
<!-- Provide a quick summary of what the model is/does. -->
|
9 |
|
10 |
-
|
11 |
-
|
12 |
## Model Details
|
13 |
|
14 |
### Model Description
|
15 |
|
16 |
<!-- Provide a longer summary of what this model is. -->
|
|
|
17 |
|
18 |
-
This is the model card of a π€ transformers model that has been pushed on the Hub. This model card has been automatically generated.
|
19 |
|
20 |
-
- **Developed by:** [
|
21 |
-
- **
|
22 |
-
- **
|
23 |
-
- **
|
24 |
-
- **
|
25 |
-
- **License:** [More Information Needed]
|
26 |
-
- **Finetuned from model [optional]:** [More Information Needed]
|
27 |
|
28 |
### Model Sources [optional]
|
29 |
|
30 |
<!-- Provide the basic links for the model. -->
|
31 |
-
|
32 |
-
- **Repository:** [More Information Needed]
|
33 |
-
- **Paper [optional]:** [More Information Needed]
|
34 |
-
- **Demo [optional]:** [More Information Needed]
|
35 |
|
36 |
## Uses
|
37 |
|
38 |
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
43 |
-
|
44 |
-
[More Information Needed]
|
45 |
-
|
46 |
-
### Downstream Use [optional]
|
47 |
-
|
48 |
-
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
49 |
-
|
50 |
-
[More Information Needed]
|
51 |
-
|
52 |
-
### Out-of-Scope Use
|
53 |
-
|
54 |
-
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
55 |
-
|
56 |
-
[More Information Needed]
|
57 |
|
58 |
## Bias, Risks, and Limitations
|
59 |
|
60 |
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
61 |
|
62 |
-
|
63 |
-
|
64 |
-
### Recommendations
|
65 |
-
|
66 |
-
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
67 |
-
|
68 |
-
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
69 |
|
70 |
## How to Get Started with the Model
|
71 |
|
72 |
Use the code below to get started with the model.
|
73 |
|
74 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
75 |
|
76 |
## Training Details
|
77 |
|
@@ -79,121 +89,193 @@ Use the code below to get started with the model.
|
|
79 |
|
80 |
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
81 |
|
82 |
-
|
83 |
-
|
84 |
-
### Training Procedure
|
85 |
-
|
86 |
-
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
87 |
-
|
88 |
-
#### Preprocessing [optional]
|
89 |
-
|
90 |
-
[More Information Needed]
|
91 |
-
|
92 |
-
|
93 |
-
#### Training Hyperparameters
|
94 |
-
|
95 |
-
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
96 |
-
|
97 |
-
#### Speeds, Sizes, Times [optional]
|
98 |
-
|
99 |
-
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
100 |
-
|
101 |
-
[More Information Needed]
|
102 |
-
|
103 |
-
## Evaluation
|
104 |
-
|
105 |
-
<!-- This section describes the evaluation protocols and provides the results. -->
|
106 |
-
|
107 |
-
### Testing Data, Factors & Metrics
|
108 |
-
|
109 |
-
#### Testing Data
|
110 |
-
|
111 |
-
<!-- This should link to a Dataset Card if possible. -->
|
112 |
-
|
113 |
-
[More Information Needed]
|
114 |
-
|
115 |
-
#### Factors
|
116 |
-
|
117 |
-
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
118 |
|
119 |
-
[
|
120 |
|
121 |
-
|
|
|
|
|
|
|
|
|
122 |
|
123 |
-
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
124 |
|
125 |
-
[
|
|
|
|
|
|
|
126 |
|
127 |
-
|
128 |
|
129 |
-
|
130 |
-
|
131 |
-
#### Summary
|
132 |
-
|
133 |
-
|
134 |
-
|
135 |
-
## Model Examination [optional]
|
136 |
-
|
137 |
-
<!-- Relevant interpretability work for the model goes here -->
|
138 |
-
|
139 |
-
[More Information Needed]
|
140 |
-
|
141 |
-
## Environmental Impact
|
142 |
-
|
143 |
-
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
144 |
-
|
145 |
-
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
146 |
-
|
147 |
-
- **Hardware Type:** [More Information Needed]
|
148 |
-
- **Hours used:** [More Information Needed]
|
149 |
-
- **Cloud Provider:** [More Information Needed]
|
150 |
-
- **Compute Region:** [More Information Needed]
|
151 |
-
- **Carbon Emitted:** [More Information Needed]
|
152 |
-
|
153 |
-
## Technical Specifications [optional]
|
154 |
-
|
155 |
-
### Model Architecture and Objective
|
156 |
-
|
157 |
-
[More Information Needed]
|
158 |
-
|
159 |
-
### Compute Infrastructure
|
160 |
-
|
161 |
-
[More Information Needed]
|
162 |
-
|
163 |
-
#### Hardware
|
164 |
-
|
165 |
-
[More Information Needed]
|
166 |
-
|
167 |
-
#### Software
|
168 |
-
|
169 |
-
[More Information Needed]
|
170 |
-
|
171 |
-
## Citation [optional]
|
172 |
-
|
173 |
-
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
174 |
-
|
175 |
-
**BibTeX:**
|
176 |
-
|
177 |
-
[More Information Needed]
|
178 |
-
|
179 |
-
**APA:**
|
180 |
-
|
181 |
-
[More Information Needed]
|
182 |
-
|
183 |
-
## Glossary [optional]
|
184 |
-
|
185 |
-
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
186 |
-
|
187 |
-
[More Information Needed]
|
188 |
-
|
189 |
-
## More Information [optional]
|
190 |
-
|
191 |
-
[More Information Needed]
|
192 |
-
|
193 |
-
## Model Card Authors [optional]
|
194 |
-
|
195 |
-
[More Information Needed]
|
196 |
|
197 |
-
|
|
|
198 |
|
199 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
+
license: mit
|
3 |
+
datasets:
|
4 |
+
- philipp-zettl/long-qa
|
5 |
+
language:
|
6 |
+
- en
|
7 |
library_name: transformers
|
8 |
+
pipeline_tag: text2text-generation
|
9 |
+
widget:
|
10 |
+
- text: "question: What's part of the Hugging Face Hub? context: The Hugging Face Hub is a
|
11 |
+
platform with over 350k models, 75k datasets, and 150k demo apps (Spaces),
|
12 |
+
all open source and publicly available, in an online platform where people
|
13 |
+
can easily collaborate and build ML together. The Hub works as a central
|
14 |
+
place where anyone can explore, experiment, collaborate, and build
|
15 |
+
technology with Machine Learning. Are you ready to join the path towards
|
16 |
+
open source Machine Learning? π€"
|
17 |
+
example_title: π€ Hub
|
18 |
+
- text: "question: What data sets can be accessed in the Datasets library? context:
|
19 |
+
π€ Datasets is a library for easily accessing and sharing datasets for Audio,
|
20 |
+
Computer Vision, and Natural Language Processing (NLP) tasks. Load a dataset
|
21 |
+
in a single line of code, and use our powerful data processing methods to
|
22 |
+
quickly get your dataset ready for training in a deep learning model. Backed
|
23 |
+
by the Apache Arrow format, process large datasets with zero-copy reads without
|
24 |
+
any memory constraints for optimal speed and efficiency. We also feature a
|
25 |
+
deep integration with the Hugging Face Hub, allowing you to easily load
|
26 |
+
and share a dataset with the wider machine learning community. Find your
|
27 |
+
dataset today on the Hugging Face Hub, and take an in-depth look inside of
|
28 |
+
it with the live viewer."
|
29 |
+
example_title: π€ datasets
|
30 |
---
|
31 |
|
32 |
+
# Model Card for t5-small-long-qa
|
33 |
|
34 |
<!-- Provide a quick summary of what the model is/does. -->
|
35 |
|
|
|
|
|
36 |
## Model Details
|
37 |
|
38 |
### Model Description
|
39 |
|
40 |
<!-- Provide a longer summary of what this model is. -->
|
41 |
+
This model was trained to generate answers for questions out of a given context.
|
42 |
|
|
|
43 |
|
44 |
+
- **Developed by:** [philipp-zettl](https://huggingface.co/philipp-zettl)
|
45 |
+
- **Model type:** Transformer (T5)
|
46 |
+
- **Language(s) (NLP):** English
|
47 |
+
- **License:** M.I.T
|
48 |
+
- **Finetuned from model [optional]:** [google/flan-t5-small](https://huggingface.co/google/flan-t5-small)
|
|
|
|
|
49 |
|
50 |
### Model Sources [optional]
|
51 |
|
52 |
<!-- Provide the basic links for the model. -->
|
53 |
+
Fine-tune of the amazing [google/flan-t5-small](https://huggingface.co/google/flan-t5-small)
|
|
|
|
|
|
|
54 |
|
55 |
## Uses
|
56 |
|
57 |
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
58 |
+
It's intended to use the model to answers for questions from given context.
|
59 |
+
The context should not exceed the model's _context_ length.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
60 |
|
61 |
## Bias, Risks, and Limitations
|
62 |
|
63 |
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
64 |
|
65 |
+
No bias evaluation was performed on this model.
|
|
|
|
|
|
|
|
|
|
|
|
|
66 |
|
67 |
## How to Get Started with the Model
|
68 |
|
69 |
Use the code below to get started with the model.
|
70 |
|
71 |
+
```python
|
72 |
+
context = "This is a long text based of multiple concatenated paragraphs."
|
73 |
+
question = "My question about something mentioned inside the context."
|
74 |
+
|
75 |
+
model_inputs = tokenizer([f"question: {question} context: {context}"], max_length=512, padding=True, truncation=True)
|
76 |
+
input_ids = torch.tensor(model_inputs['input_ids']).to(device)
|
77 |
+
attention_mask = torch.tensor(model_inputs['attention_mask']).to(device)
|
78 |
+
with torch.no_grad():
|
79 |
+
sample_output = model.generate(input_ids[:1], max_length=85)
|
80 |
+
sample_output_text = tokenizer.decode(sample_output[0], skip_special_tokens=True)
|
81 |
+
input_text = tokenizer.decode(input_ids[0], skip_special_tokens=True)
|
82 |
+
print(f"Sample Input:\n \"{input_text}\"\n\n")
|
83 |
+
print(f"Model Output: \"{sample_output_text}\"")
|
84 |
+
```
|
85 |
|
86 |
## Training Details
|
87 |
|
|
|
89 |
|
90 |
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
91 |
|
92 |
+
This model was trained on [philipp-zettl/long-qa](https://huggingface.co/datasets/philipp-zettl/long-qa).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
93 |
|
94 |
+
A synthetic data set created from [philipp-zettl/qg-tidyqa_squad2](https://huggingface.co/datasets/philipp-zettl/qg-tydiqa_squad2) using [microsoft/Phi-3-mini-128k-instruct](https://huggingface.co/microsoft/Phi-3-mini-128k-instruct).
|
95 |
|
96 |
+
The data set was created by prompting Phi-3 using the prompt template
|
97 |
+
```python
|
98 |
+
msg = f"""
|
99 |
+
Answer the following question using the content provided in the context.
|
100 |
+
Do not answer questions where the answer isn't inside the context.
|
101 |
|
|
|
102 |
|
103 |
+
Question: {sample['question']}
|
104 |
+
Context: {sample['context']}
|
105 |
+
"""
|
106 |
+
```
|
107 |
|
108 |
+
After generating synthetic answers, the data set was manually corrected and validated to ensure high quality as well as consistent longer answers than the original data sets.
|
109 |
|
110 |
+
### Training Procedure
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
111 |
|
112 |
+
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
113 |
+
Below you can find the full training pipeline used to achieve this fine-tune.
|
114 |
|
115 |
+
```python
|
116 |
+
import torch
|
117 |
+
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
|
118 |
+
|
119 |
+
# Base model (e.g., T5-large)
|
120 |
+
# https://huggingface.co/collections/google/flan-t5-release-65005c39e3201fff885e22fb
|
121 |
+
model_name = 'google/flan-t5-small'
|
122 |
+
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
|
123 |
+
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
124 |
+
|
125 |
+
# Move only the student model to GPU if available
|
126 |
+
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
|
127 |
+
model = model.to(device)
|
128 |
+
```
|
129 |
+
|
130 |
+
Load dataset
|
131 |
+
```python
|
132 |
+
from datasets import load_dataset
|
133 |
+
|
134 |
+
# Load dataset
|
135 |
+
ds = load_dataset('philipp-zettl/long-qa')
|
136 |
+
|
137 |
+
# Split the dataset into training and validation
|
138 |
+
train_dataset = ds['train']
|
139 |
+
validation_dataset = ds['test']
|
140 |
+
```
|
141 |
+
|
142 |
+
Preprocessing: tokenize inputs and labels for faster training cycles, i.e. no need for tokenization during training anymore
|
143 |
+
```python
|
144 |
+
def preprocess_batch(batch, tokenizer, max_input_length=512, max_output_length=128):
|
145 |
+
questions = batch['question']
|
146 |
+
contexts = batch['context']
|
147 |
+
answers = batch['answer']
|
148 |
+
|
149 |
+
inputs = [f"question: {q} context: {c}" for q, c in zip(questions, contexts)]
|
150 |
+
model_inputs = tokenizer(inputs, max_length=max_input_length, padding=True, truncation=True)
|
151 |
+
|
152 |
+
labels = tokenizer(answers, max_length=max_output_length, padding=True, truncation=True)
|
153 |
+
model_inputs['labels'] = labels['input_ids']
|
154 |
+
|
155 |
+
return model_inputs
|
156 |
+
|
157 |
+
# Tokenize the dataset
|
158 |
+
train_dataset = train_dataset.map(lambda batch: preprocess_batch(batch, tokenizer), batched=True)
|
159 |
+
validation_dataset = validation_dataset.map(lambda batch: preprocess_batch(batch, tokenizer), batched=True)
|
160 |
+
|
161 |
+
# Set format for PyTorch
|
162 |
+
train_dataset.set_format(type='torch', columns=['input_ids', 'attention_mask', 'labels'])
|
163 |
+
validation_dataset.set_format(type='torch', columns=['input_ids', 'attention_mask', 'labels'])
|
164 |
+
```
|
165 |
+
|
166 |
+
The train loop
|
167 |
+
```python
|
168 |
+
from tqdm import tqdm
|
169 |
+
from transformers import AdamW, DataCollatorForSeq2Seq
|
170 |
+
from torch.utils.data import DataLoader
|
171 |
+
from torch.utils.tensorboard import SummaryWriter
|
172 |
+
|
173 |
+
torch.cuda.empty_cache()
|
174 |
+
|
175 |
+
model_name = 'google/flan-t5-small'
|
176 |
+
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
|
177 |
+
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
178 |
+
|
179 |
+
# Move only the student model to GPU if available
|
180 |
+
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
|
181 |
+
model = model.to(device)
|
182 |
+
|
183 |
+
# Training parameters
|
184 |
+
epochs = 50
|
185 |
+
learning_rate = 3e-5
|
186 |
+
temperature = 2.0
|
187 |
+
batch_size = 8
|
188 |
+
optimizer = torch.optim.AdamW(model.parameters(), lr=learning_rate)
|
189 |
+
|
190 |
+
# Create a data collator for padding and batching
|
191 |
+
data_collator = DataCollatorForSeq2Seq(tokenizer=tokenizer, model=model)
|
192 |
+
|
193 |
+
# Create DataLoaders with the data collator
|
194 |
+
train_dataloader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True, collate_fn=data_collator)
|
195 |
+
validation_dataloader = DataLoader(validation_dataset, batch_size=batch_size, collate_fn=data_collator)
|
196 |
+
|
197 |
+
writer = SummaryWriter(comment='t5-small-long-qa')
|
198 |
+
|
199 |
+
# Store losses and learning rates
|
200 |
+
train_losses = []
|
201 |
+
val_losses = []
|
202 |
+
learning_rates = []
|
203 |
+
|
204 |
+
print("Starting training...")
|
205 |
+
|
206 |
+
# Training loop
|
207 |
+
for epoch in range(epochs):
|
208 |
+
model.train()
|
209 |
+
total_loss = 0
|
210 |
+
print(f"Epoch {epoch+1}/{epochs}")
|
211 |
+
|
212 |
+
progress_bar = tqdm(train_dataloader, desc="Training", leave=False)
|
213 |
+
|
214 |
+
for step, batch in enumerate(progress_bar):
|
215 |
+
# Move student inputs to GPU
|
216 |
+
input_ids = batch['input_ids'].to(device)
|
217 |
+
attention_mask = batch['attention_mask'].to(device)
|
218 |
+
labels = batch['labels'].to(device)
|
219 |
+
|
220 |
+
# Teacher forward pass on CPU
|
221 |
+
outputs = model(input_ids=input_ids, attention_mask=attention_mask, labels=labels)
|
222 |
+
logits = outputs.logits
|
223 |
+
|
224 |
+
# Calculate losses
|
225 |
+
loss = outputs.loss # Cross-entropy loss
|
226 |
+
writer.add_scalar("Loss/train", loss, epoch * len(train_dataloader) + step)
|
227 |
+
|
228 |
+
# Backpropagation
|
229 |
+
optimizer.zero_grad()
|
230 |
+
loss.backward()
|
231 |
+
optimizer.step()
|
232 |
+
|
233 |
+
total_loss += loss.item()
|
234 |
+
|
235 |
+
# Verbose logging
|
236 |
+
if step % len(train_dataloader)//10 == 1 or step == len(train_dataloader) - 1:
|
237 |
+
progress_bar.set_postfix({
|
238 |
+
'step': step,
|
239 |
+
'loss': loss.item(),
|
240 |
+
})
|
241 |
+
|
242 |
+
# Generate a sample output from the student model
|
243 |
+
model.eval()
|
244 |
+
with torch.no_grad():
|
245 |
+
sample_output = model.generate(input_ids[:1], max_length=50)
|
246 |
+
sample_output_text = tokenizer.decode(sample_output[0], skip_special_tokens=True)
|
247 |
+
input_text = tokenizer.decode(input_ids[0], skip_special_tokens=True)
|
248 |
+
writer.add_text(f"Sample Input", input_text, step)
|
249 |
+
writer.add_text(f"Sample Output", sample_output_text, step)
|
250 |
+
model.train()
|
251 |
+
|
252 |
+
|
253 |
+
avg_train_loss = total_loss / len(train_dataloader)
|
254 |
+
train_losses.append(avg_train_loss)
|
255 |
+
learning_rates.append(optimizer.param_groups[0]['lr'])
|
256 |
+
|
257 |
+
# Validation step
|
258 |
+
model.eval()
|
259 |
+
total_val_loss = 0
|
260 |
+
with torch.no_grad():
|
261 |
+
for batch in validation_dataloader:
|
262 |
+
input_ids = batch['input_ids'].to(device)
|
263 |
+
attention_mask = batch['attention_mask'].to(device)
|
264 |
+
labels = batch['labels'].to(device)
|
265 |
+
|
266 |
+
outputs = model(input_ids=input_ids, attention_mask=attention_mask, labels=labels)
|
267 |
+
val_loss = outputs.loss
|
268 |
+
total_val_loss += val_loss.item()
|
269 |
+
|
270 |
+
avg_val_loss = total_val_loss / len(validation_dataloader)
|
271 |
+
val_losses.append(avg_val_loss)
|
272 |
+
|
273 |
+
writer.add_scalar("AVG Loss/train", avg_train_loss, epoch)
|
274 |
+
writer.add_scalar("AVG Loss/val", avg_val_loss, epoch)
|
275 |
+
|
276 |
+
print(f"Epoch {epoch+1} completed. Avg Train Loss: {avg_train_loss:.4f}, Avg Val Loss: {avg_val_loss:.4f}")
|
277 |
+
|
278 |
+
|
279 |
+
print("Training complete.")
|
280 |
+
writer.close()
|
281 |
+
```
|