upload model
Browse files- .gitattributes +1 -0
- README.md +56 -3
- config.json +32 -0
- generation_config.json +7 -0
- pytorch_model.bin +3 -0
- special_tokens_map.json +6 -0
- tokenizer.json +3 -0
- tokenizer_config.json +11 -0
.gitattributes
CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
36 |
+
tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
README.md
CHANGED
@@ -1,3 +1,56 @@
|
|
1 |
-
|
2 |
-
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# RankingGPT-bloom-560m
|
2 |
+
|
3 |
+
RankingGPT is a text ranker based on large language models with significant in-domain and out-domain effectiveness.
|
4 |
+
We provide RankingGPT in different sizes and types, including bloom-560m, bloom-1b1, bloom-3b, bloom-7b, llama2-7b, baichuan2-7b and qwen-7b.
|
5 |
+
|
6 |
+
More details please refer to our [paper](https://arxiv.org/abs/2311.16720) and [github](https://github.com/Alibaba-NLP/RankingGPT).
|
7 |
+
|
8 |
+
|
9 |
+
## Usage
|
10 |
+
|
11 |
+
Code example
|
12 |
+
```python
|
13 |
+
import torch
|
14 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM
|
15 |
+
|
16 |
+
tokenizer = AutoTokenizer.from_pretrained('RankingGPT-bloom-560m')
|
17 |
+
model = AutoModelForCausalLM.from_pretrained('RankingGPT-bloom-560m').eval()
|
18 |
+
|
19 |
+
query='when should a baby walk'
|
20 |
+
document='Most babies start to walk around 13 months, but your baby may start walking as early as 9 or 10 months or as late as 15 or 16 months.'
|
21 |
+
|
22 |
+
context=f'Document: {document} Query:'
|
23 |
+
example=context+query
|
24 |
+
|
25 |
+
context_enc = tokenizer.encode(context, add_special_tokens=False)
|
26 |
+
continuation_enc = tokenizer.encode(query, add_special_tokens=False)
|
27 |
+
model_input = torch.tensor(context_enc+continuation_enc[:-1])
|
28 |
+
continuation_len = len(continuation_enc)
|
29 |
+
input_len, = model_input.shape
|
30 |
+
|
31 |
+
|
32 |
+
with torch.no_grad():
|
33 |
+
logprobs = torch.nn.functional.log_softmax(model(model_input.unsqueeze(dim=0))[0], dim=-1)[0]
|
34 |
+
|
35 |
+
logprobs = logprobs[input_len-continuation_len:]
|
36 |
+
logprobs = torch.gather(logprobs, 1, torch.tensor(continuation_enc).unsqueeze(-1)).squeeze(-1)
|
37 |
+
score = torch.sum(logprobs)/logprobs.shape[0]
|
38 |
+
|
39 |
+
print(f"Document: {document[:20] + '...'} Score: {score}")
|
40 |
+
```
|
41 |
+
|
42 |
+
|
43 |
+
### Citation
|
44 |
+
|
45 |
+
If you find our paper or models helpful, please consider citing them as follows:
|
46 |
+
|
47 |
+
```
|
48 |
+
@misc{zhang2023rankinggpt,
|
49 |
+
title={RankingGPT: Empowering Large Language Models in Text Ranking with Progressive Enhancement},
|
50 |
+
author={Longhui Zhang and Yanzhao Zhang and Dingkun Long and Pengjun Xie and Meishan Zhang and Min Zhang},
|
51 |
+
year={2023},
|
52 |
+
eprint={2311.16720},
|
53 |
+
archivePrefix={arXiv},
|
54 |
+
primaryClass={cs.IR}
|
55 |
+
}
|
56 |
+
```
|
config.json
ADDED
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "bigscience/bloom-560m",
|
3 |
+
"apply_residual_connection_post_layernorm": false,
|
4 |
+
"architectures": [
|
5 |
+
"BloomForCausalLM"
|
6 |
+
],
|
7 |
+
"attention_dropout": 0.0,
|
8 |
+
"attention_softmax_in_fp32": true,
|
9 |
+
"bias_dropout_fusion": true,
|
10 |
+
"bos_token_id": 1,
|
11 |
+
"eos_token_id": 2,
|
12 |
+
"hidden_dropout": 0.0,
|
13 |
+
"hidden_size": 1024,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"layer_norm_epsilon": 1e-05,
|
16 |
+
"masked_softmax_fusion": true,
|
17 |
+
"model_type": "bloom",
|
18 |
+
"n_head": 16,
|
19 |
+
"n_inner": null,
|
20 |
+
"n_layer": 24,
|
21 |
+
"offset_alibi": 100,
|
22 |
+
"pad_token_id": 3,
|
23 |
+
"pretraining_tp": 1,
|
24 |
+
"skip_bias_add": true,
|
25 |
+
"skip_bias_add_qkv": false,
|
26 |
+
"slow_but_exact": false,
|
27 |
+
"torch_dtype": "float32",
|
28 |
+
"transformers_version": "4.29.0",
|
29 |
+
"unk_token_id": 0,
|
30 |
+
"use_cache": true,
|
31 |
+
"vocab_size": 250880
|
32 |
+
}
|
generation_config.json
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_from_model_config": true,
|
3 |
+
"bos_token_id": 1,
|
4 |
+
"eos_token_id": 2,
|
5 |
+
"pad_token_id": 3,
|
6 |
+
"transformers_version": "4.29.0"
|
7 |
+
}
|
pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:daff937349f9c8bbc30bb810544328c0fa1bb4ba95537c32d08b3696e744d2fd
|
3 |
+
size 2236951031
|
special_tokens_map.json
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"bos_token": "<s>",
|
3 |
+
"eos_token": "</s>",
|
4 |
+
"pad_token": "<pad>",
|
5 |
+
"unk_token": "<unk>"
|
6 |
+
}
|
tokenizer.json
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8f6efc66e73f1fd69da4f436e48befb519fdff3fe18910850c1d41bd862293a5
|
3 |
+
size 14500443
|
tokenizer_config.json
ADDED
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"add_prefix_space": false,
|
3 |
+
"bos_token": "<s>",
|
4 |
+
"clean_up_tokenization_spaces": false,
|
5 |
+
"eos_token": "</s>",
|
6 |
+
"model_max_length": 512,
|
7 |
+
"pad_token": "<pad>",
|
8 |
+
"padding_side": "right",
|
9 |
+
"tokenizer_class": "BloomTokenizer",
|
10 |
+
"unk_token": "<unk>"
|
11 |
+
}
|