julianrisch
commited on
Commit
•
eea39c6
1
Parent(s):
ec78f8d
Update README.md
Browse files
README.md
CHANGED
@@ -140,7 +140,7 @@ model-index:
|
|
140 |
name: F1
|
141 |
---
|
142 |
|
143 |
-
# deberta-v3-base for QA
|
144 |
|
145 |
This is the [deberta-v3-base](https://huggingface.co/microsoft/deberta-v3-base) model, fine-tuned using the [SQuAD2.0](https://huggingface.co/datasets/squad_v2) dataset. It's been trained on question-answer pairs, including unanswerable questions, for the task of Question Answering.
|
146 |
|
@@ -151,7 +151,7 @@ This is the [deberta-v3-base](https://huggingface.co/microsoft/deberta-v3-base)
|
|
151 |
**Downstream-task:** Extractive QA
|
152 |
**Training data:** SQuAD 2.0
|
153 |
**Eval data:** SQuAD 2.0
|
154 |
-
**Code:** See [an example QA pipeline
|
155 |
**Infrastructure**: 1x NVIDIA A10G
|
156 |
|
157 |
## Hyperparameters
|
@@ -171,17 +171,34 @@ max_query_length = 64
|
|
171 |
## Usage
|
172 |
|
173 |
### In Haystack
|
174 |
-
Haystack is an
|
|
|
175 |
```python
|
176 |
-
|
177 |
-
|
178 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
179 |
```
|
|
|
180 |
|
181 |
### In Transformers
|
182 |
```python
|
183 |
from transformers import AutoModelForQuestionAnswering, AutoTokenizer, pipeline
|
184 |
-
|
|
|
|
|
185 |
# a) Get predictions
|
186 |
nlp = pipeline('question-answering', model=model_name, tokenizer=model_name)
|
187 |
QA_input = {
|
@@ -189,38 +206,42 @@ QA_input = {
|
|
189 |
'context': 'The option to convert models between FARM and transformers gives freedom to the user and let people easily switch between frameworks.'
|
190 |
}
|
191 |
res = nlp(QA_input)
|
|
|
192 |
# b) Load model & tokenizer
|
193 |
model = AutoModelForQuestionAnswering.from_pretrained(model_name)
|
194 |
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
195 |
```
|
196 |
|
|
|
197 |
## Authors
|
198 |
**Sebastian Lee:** sebastian.lee [at] deepset.ai
|
199 |
**Timo Möller:** timo.moeller [at] deepset.ai
|
200 |
**Malte Pietsch:** malte.pietsch [at] deepset.ai
|
201 |
|
202 |
## About us
|
|
|
203 |
<div class="grid lg:grid-cols-2 gap-x-4 gap-y-3">
|
204 |
<div class="w-full h-40 object-cover mb-2 rounded-lg flex items-center justify-center">
|
205 |
-
<img alt="" src="https://
|
206 |
</div>
|
207 |
-
|
208 |
-
<img alt="" src="https://
|
209 |
</div>
|
210 |
</div>
|
211 |
|
212 |
-
[deepset](http://deepset.ai/) is the company behind the open-source
|
213 |
-
|
214 |
|
215 |
Some of our other work:
|
216 |
-
- [Distilled roberta-base-squad2 (aka "tinyroberta-squad2")](
|
217 |
-
- [German BERT
|
218 |
-
- [
|
219 |
|
220 |
## Get in touch and join the Haystack community
|
221 |
|
222 |
-
<p>For more info on Haystack, visit our <strong><a href="https://github.com/deepset-ai/haystack">GitHub</a></strong> repo and <strong><a href="https://haystack.deepset.ai">Documentation</a></strong>.
|
|
|
|
|
223 |
|
224 |
-
|
225 |
|
226 |
-
|
|
|
140 |
name: F1
|
141 |
---
|
142 |
|
143 |
+
# deberta-v3-base for Extractive QA
|
144 |
|
145 |
This is the [deberta-v3-base](https://huggingface.co/microsoft/deberta-v3-base) model, fine-tuned using the [SQuAD2.0](https://huggingface.co/datasets/squad_v2) dataset. It's been trained on question-answer pairs, including unanswerable questions, for the task of Question Answering.
|
146 |
|
|
|
151 |
**Downstream-task:** Extractive QA
|
152 |
**Training data:** SQuAD 2.0
|
153 |
**Eval data:** SQuAD 2.0
|
154 |
+
**Code:** See [an example extractive QA pipeline built with Haystack](https://haystack.deepset.ai/tutorials/34_extractive_qa_pipeline)
|
155 |
**Infrastructure**: 1x NVIDIA A10G
|
156 |
|
157 |
## Hyperparameters
|
|
|
171 |
## Usage
|
172 |
|
173 |
### In Haystack
|
174 |
+
Haystack is an AI orchestration framework to build customizable, production-ready LLM applications. You can use this model in Haystack to do extractive question answering on documents.
|
175 |
+
To load and run the model with [Haystack](https://github.com/deepset-ai/haystack/):
|
176 |
```python
|
177 |
+
# After running pip install haystack-ai "transformers[torch,sentencepiece]"
|
178 |
+
|
179 |
+
from haystack import Document
|
180 |
+
from haystack.components.readers import ExtractiveReader
|
181 |
+
|
182 |
+
docs = [
|
183 |
+
Document(content="Python is a popular programming language"),
|
184 |
+
Document(content="python ist eine beliebte Programmiersprache"),
|
185 |
+
]
|
186 |
+
|
187 |
+
reader = ExtractiveReader(model="deepset/roberta-base-squad2")
|
188 |
+
reader.warm_up()
|
189 |
+
|
190 |
+
question = "What is a popular programming language?"
|
191 |
+
result = reader.run(query=question, documents=docs)
|
192 |
+
# {'answers': [ExtractedAnswer(query='What is a popular programming language?', score=0.5740374326705933, data='python', document=Document(id=..., content: '...'), context=None, document_offset=ExtractedAnswer.Span(start=0, end=6),...)]}
|
193 |
```
|
194 |
+
For a complete example with an extractive question answering pipeline that scales over many documents, check out the [corresponding Haystack tutorial](https://haystack.deepset.ai/tutorials/34_extractive_qa_pipeline).
|
195 |
|
196 |
### In Transformers
|
197 |
```python
|
198 |
from transformers import AutoModelForQuestionAnswering, AutoTokenizer, pipeline
|
199 |
+
|
200 |
+
model_name = "deepset/roberta-base-squad2"
|
201 |
+
|
202 |
# a) Get predictions
|
203 |
nlp = pipeline('question-answering', model=model_name, tokenizer=model_name)
|
204 |
QA_input = {
|
|
|
206 |
'context': 'The option to convert models between FARM and transformers gives freedom to the user and let people easily switch between frameworks.'
|
207 |
}
|
208 |
res = nlp(QA_input)
|
209 |
+
|
210 |
# b) Load model & tokenizer
|
211 |
model = AutoModelForQuestionAnswering.from_pretrained(model_name)
|
212 |
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
213 |
```
|
214 |
|
215 |
+
|
216 |
## Authors
|
217 |
**Sebastian Lee:** sebastian.lee [at] deepset.ai
|
218 |
**Timo Möller:** timo.moeller [at] deepset.ai
|
219 |
**Malte Pietsch:** malte.pietsch [at] deepset.ai
|
220 |
|
221 |
## About us
|
222 |
+
|
223 |
<div class="grid lg:grid-cols-2 gap-x-4 gap-y-3">
|
224 |
<div class="w-full h-40 object-cover mb-2 rounded-lg flex items-center justify-center">
|
225 |
+
<img alt="" src="https://raw.githubusercontent.com/deepset-ai/.github/main/deepset-logo-colored.png" class="w-40"/>
|
226 |
</div>
|
227 |
+
<div class="w-full h-40 object-cover mb-2 rounded-lg flex items-center justify-center">
|
228 |
+
<img alt="" src="https://raw.githubusercontent.com/deepset-ai/.github/main/haystack-logo-colored.png" class="w-40"/>
|
229 |
</div>
|
230 |
</div>
|
231 |
|
232 |
+
[deepset](http://deepset.ai/) is the company behind the production-ready open-source AI framework [Haystack](https://haystack.deepset.ai/).
|
|
|
233 |
|
234 |
Some of our other work:
|
235 |
+
- [Distilled roberta-base-squad2 (aka "tinyroberta-squad2")](https://huggingface.co/deepset/tinyroberta-squad2)
|
236 |
+
- [German BERT](https://deepset.ai/german-bert), [GermanQuAD and GermanDPR](https://deepset.ai/germanquad), [German embedding model](https://huggingface.co/mixedbread-ai/deepset-mxbai-embed-de-large-v1)
|
237 |
+
- [deepset Cloud](https://www.deepset.ai/deepset-cloud-product), [deepset Studio](https://www.deepset.ai/deepset-studio)
|
238 |
|
239 |
## Get in touch and join the Haystack community
|
240 |
|
241 |
+
<p>For more info on Haystack, visit our <strong><a href="https://github.com/deepset-ai/haystack">GitHub</a></strong> repo and <strong><a href="https://docs.haystack.deepset.ai">Documentation</a></strong>.
|
242 |
+
|
243 |
+
We also have a <strong><a class="h-7" href="https://haystack.deepset.ai/community">Discord community open to everyone!</a></strong></p>
|
244 |
|
245 |
+
[Twitter](https://twitter.com/Haystack_AI) | [LinkedIn](https://www.linkedin.com/company/deepset-ai/) | [Discord](https://haystack.deepset.ai/community) | [GitHub Discussions](https://github.com/deepset-ai/haystack/discussions) | [Website](https://haystack.deepset.ai/) | [YouTube](https://www.youtube.com/@deepset_ai)
|
246 |
|
247 |
+
By the way: [we're hiring!](http://www.deepset.ai/jobs)
|