Update README.md
Browse files
README.md
CHANGED
@@ -21,7 +21,8 @@ We introduce GNER, a **G**enerative **N**amed **E**ntity **R**ecognition framewo
|
|
21 |
* π» Code: [https://github.com/yyDing1/GNER/](https://github.com/yyDing1/GNER/)
|
22 |
* π Paper: [Rethinking Negative Instances for Generative Named Entity Recognition](https://arxiv.org/abs/2402.16602)
|
23 |
* πΎ Models in the π€ HuggingFace Hub: [GNER-Models](https://huggingface.co/collections/dyyyyyyyy/gner-65dda2cb96c6e35c814dea56)
|
24 |
-
*
|
|
|
25 |
|
26 |
<p align="center">
|
27 |
<img src="https://github.com/yyDing1/GNER/raw/main/assets/zero_shot_results.png">
|
@@ -49,9 +50,9 @@ pip install torch>=2.1.0 datasets>=2.17.0 deepspeed>=0.13.4 accelerate>=0.27.2 t
|
|
49 |
Below is an example using `GNER-LLaMA`
|
50 |
```python
|
51 |
>>> import torch
|
52 |
-
>>> from transformers import AutoTokenizer,
|
53 |
>>> tokenizer = AutoTokenizer.from_pretrained("dyyyyyyyy/GNER-LLaMA-7B")
|
54 |
-
>>> model =
|
55 |
>>> model = model.eval()
|
56 |
>>> instruction_template = "Please analyze the sentence provided, identifying the type of entity for each word on a token-by-token basis.\nOutput format is: word_1(label_1), word_2(label_2), ...\nWe'll use the BIO-format to label the entities, where:\n1. B- (Begin) indicates the start of a named entity.\n2. I- (Inside) is used for words within a named entity but are not the first word.\n3. O (Outside) denotes words that are not part of a named entity.\n"
|
57 |
>>> sentence = "did george clooney make a musical in the 1980s"
|
@@ -61,7 +62,7 @@ Below is an example using `GNER-LLaMA`
|
|
61 |
>>> inputs = tokenizer(instruction, return_tensors="pt").to("cuda")
|
62 |
>>> outputs = model.generate(**inputs, max_new_tokens=640)
|
63 |
>>> response = tokenizer.decode(outputs[0], skip_special_tokens=True)
|
64 |
-
>>> response = response[
|
65 |
>>> print(response)
|
66 |
"did(O) george(B-actor) clooney(I-actor) make(O) a(O) musical(B-genre) in(O) the(O) 1980s(B-year)"
|
67 |
```
|
|
|
21 |
* π» Code: [https://github.com/yyDing1/GNER/](https://github.com/yyDing1/GNER/)
|
22 |
* π Paper: [Rethinking Negative Instances for Generative Named Entity Recognition](https://arxiv.org/abs/2402.16602)
|
23 |
* πΎ Models in the π€ HuggingFace Hub: [GNER-Models](https://huggingface.co/collections/dyyyyyyyy/gner-65dda2cb96c6e35c814dea56)
|
24 |
+
* π§ͺ Reproduction Materials: [Reproduction Materials](https://drive.google.com/drive/folders/1m2FqDgItEbSoeUVo-i18AwMvBcNkZD46?usp=drive_link)
|
25 |
+
* π¨ Example Jupyter Notebooks: [GNER Notebook](https://github.com/yyDing1/GNER/blob/main/notebook.ipynb)
|
26 |
|
27 |
<p align="center">
|
28 |
<img src="https://github.com/yyDing1/GNER/raw/main/assets/zero_shot_results.png">
|
|
|
50 |
Below is an example using `GNER-LLaMA`
|
51 |
```python
|
52 |
>>> import torch
|
53 |
+
>>> from transformers import AutoTokenizer, AutoModelForCausalLM
|
54 |
>>> tokenizer = AutoTokenizer.from_pretrained("dyyyyyyyy/GNER-LLaMA-7B")
|
55 |
+
>>> model = AutoModelForCausalLM.from_pretrained("dyyyyyyyy/GNER-LLaMA-7B", torch_dtype=torch.bfloat16).cuda()
|
56 |
>>> model = model.eval()
|
57 |
>>> instruction_template = "Please analyze the sentence provided, identifying the type of entity for each word on a token-by-token basis.\nOutput format is: word_1(label_1), word_2(label_2), ...\nWe'll use the BIO-format to label the entities, where:\n1. B- (Begin) indicates the start of a named entity.\n2. I- (Inside) is used for words within a named entity but are not the first word.\n3. O (Outside) denotes words that are not part of a named entity.\n"
|
58 |
>>> sentence = "did george clooney make a musical in the 1980s"
|
|
|
62 |
>>> inputs = tokenizer(instruction, return_tensors="pt").to("cuda")
|
63 |
>>> outputs = model.generate(**inputs, max_new_tokens=640)
|
64 |
>>> response = tokenizer.decode(outputs[0], skip_special_tokens=True)
|
65 |
+
>>> response = response[response.find("[/INST]") + len("[/INST]"):].strip()
|
66 |
>>> print(response)
|
67 |
"did(O) george(B-actor) clooney(I-actor) make(O) a(O) musical(B-genre) in(O) the(O) 1980s(B-year)"
|
68 |
```
|