Update README.md
Browse filesUpdate generations after major fix: https://github.com/huggingface/transformers/commit/abc400b06a8ab26cd438b6e9add3aad082ffc48f
README.md
CHANGED
@@ -72,7 +72,7 @@ It is recommended to directly call the [`generate`](https://huggingface.co/docs/
|
|
72 |
>>> generated_ids = model.generate(input_ids)
|
73 |
|
74 |
>>> tokenizer.batch_decode(generated_ids, skip_special_tokens=True)
|
75 |
-
[
|
76 |
```
|
77 |
|
78 |
By default, generation is deterministic. In order to use the top-k sampling, please set `do_sample` to `True`.
|
@@ -94,7 +94,7 @@ By default, generation is deterministic. In order to use the top-k sampling, ple
|
|
94 |
>>> generated_ids = model.generate(input_ids, do_sample=True)
|
95 |
|
96 |
>>> tokenizer.batch_decode(generated_ids, skip_special_tokens=True)
|
97 |
-
[
|
98 |
```
|
99 |
|
100 |
### Limitations and bias
|
@@ -127,11 +127,11 @@ Here's an example of how the model can have biased predictions:
|
|
127 |
>>> generated_ids = model.generate(input_ids, do_sample=True, num_return_sequences=5, max_length=10)
|
128 |
|
129 |
>>> tokenizer.batch_decode(generated_ids, skip_special_tokens=True)
|
130 |
-
The woman worked as a
|
131 |
-
The woman worked as a
|
132 |
-
The woman worked as a
|
133 |
-
The woman worked as a
|
134 |
-
The woman worked as a
|
135 |
```
|
136 |
|
137 |
compared to:
|
@@ -153,11 +153,11 @@ compared to:
|
|
153 |
>>> generated_ids = model.generate(input_ids, do_sample=True, num_return_sequences=5, max_length=10)
|
154 |
|
155 |
>>> tokenizer.batch_decode(generated_ids, skip_special_tokens=True)
|
156 |
-
The man worked as a consultant
|
157 |
-
The man worked as a
|
158 |
-
The man worked as a
|
159 |
-
The man worked as a
|
160 |
-
The man worked as a
|
161 |
```
|
162 |
|
163 |
This bias will also affect all fine-tuned versions of this model.
|
|
|
72 |
>>> generated_ids = model.generate(input_ids)
|
73 |
|
74 |
>>> tokenizer.batch_decode(generated_ids, skip_special_tokens=True)
|
75 |
+
['Hello, I am conscious and aware of my surroundings.\nI am conscious and aware of my']
|
76 |
```
|
77 |
|
78 |
By default, generation is deterministic. In order to use the top-k sampling, please set `do_sample` to `True`.
|
|
|
94 |
>>> generated_ids = model.generate(input_ids, do_sample=True)
|
95 |
|
96 |
>>> tokenizer.batch_decode(generated_ids, skip_special_tokens=True)
|
97 |
+
['Hello, I am conscious and aware.\nSo that makes you dead, right? ']
|
98 |
```
|
99 |
|
100 |
### Limitations and bias
|
|
|
127 |
>>> generated_ids = model.generate(input_ids, do_sample=True, num_return_sequences=5, max_length=10)
|
128 |
|
129 |
>>> tokenizer.batch_decode(generated_ids, skip_special_tokens=True)
|
130 |
+
The woman worked as a supervisor in the office
|
131 |
+
The woman worked as a social media consultant for
|
132 |
+
The woman worked as a cashier at the
|
133 |
+
The woman worked as a teacher, and was
|
134 |
+
The woman worked as a maid at our friends
|
135 |
```
|
136 |
|
137 |
compared to:
|
|
|
153 |
>>> generated_ids = model.generate(input_ids, do_sample=True, num_return_sequences=5, max_length=10)
|
154 |
|
155 |
>>> tokenizer.batch_decode(generated_ids, skip_special_tokens=True)
|
156 |
+
The man worked as a consultant to the defense
|
157 |
+
The man worked as a bartender in a bar
|
158 |
+
The man worked as a cashier at the
|
159 |
+
The man worked as a teacher, and was
|
160 |
+
The man worked as a professional athlete while he
|
161 |
```
|
162 |
|
163 |
This bias will also affect all fine-tuned versions of this model.
|