clonefy commited on
Commit
b8954a7
1 Parent(s): cc5c4f7

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +8 -8
README.md CHANGED
@@ -2,7 +2,7 @@
2
  license: other
3
  license_name: tongyi-qianwen-research
4
  license_link: >-
5
- https://huggingface.co/Qwen/Qwen2-beta-4B-Chat/blob/main/LICENSE
6
  language:
7
  - en
8
  pipeline_tag: text-generation
@@ -10,12 +10,12 @@ tags:
10
  - chat
11
  ---
12
 
13
- # Qwen2-beta-4B-Chat
14
 
15
 
16
  ## Introduction
17
 
18
- Qwen2-beta is the beta version of Qwen2, a transformer-based decoder-only language model pretrained on a large amount of data. In comparison with the previous released Qwen, the improvements include:
19
 
20
  * 6 model sizes, including 0.5B, 1.8B, 4B, 7B, 14B, and 72B;
21
  * Significant performance improvement in human preference for chat models;
@@ -27,13 +27,13 @@ For more details, please refer to our blog post and GitHub repo.
27
  <br>
28
 
29
  ## Model Details
30
- Qwen2 is a language model series including decoder language models of different model sizes. For each size, we release the base language model and the aligned chat model. It is based on the Transformer architecture with SwiGLU activation, attention QKV bias, group query attention, mixture of sliding window attention and full attention, etc. Additionally, we have an improved tokenizer adaptive to multiple natural languages and codes. For the beta version, temporarily we did not include GQA and the mixture of SWA and full attention.
31
 
32
  ## Training details
33
  We pretrained the models with a large amount of data, and we post-trained the models with both supervised finetuning and direct preference optimization. However, DPO leads to improvements in human preference evaluation but degradation in benchmark evaluation. In the very near future, we will fix both problems.
34
 
35
  ## Requirements
36
- The code of Qwen2 has been in the latest Hugging face transformers and we advise you to install `transformers>=4.37.0`, or you might encounter the following error:
37
  ```
38
  KeyError: 'qwen2'
39
  ```
@@ -47,10 +47,10 @@ from transformers import AutoModelForCausalLM, AutoTokenizer
47
  device = "cuda" # the device to load the model onto
48
 
49
  model = AutoModelForCausalLM.from_pretrained(
50
- "Qwen/Qwen2-beta-4B-Chat",
51
  device_map="auto"
52
  )
53
- tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2-beta-4B-Chat")
54
 
55
  prompt = "Give me a short introduction to large language model."
56
  messages = [
@@ -75,7 +75,7 @@ generated_ids = [
75
  response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
76
  ```
77
 
78
- For quantized models, we advise you to use the GPTQ, AWQ, and GGUF correspondents, namely `Qwen-beta-4B-Chat-GPTQ`, `Qwen-beta-4B-Chat-AWQ`, and `Qwen-beta-4B-Chat-GGUF`.
79
 
80
 
81
  ## Limitations
 
2
  license: other
3
  license_name: tongyi-qianwen-research
4
  license_link: >-
5
+ https://huggingface.co/Qwen/Qwen1.5-4B-Chat/blob/main/LICENSE
6
  language:
7
  - en
8
  pipeline_tag: text-generation
 
10
  - chat
11
  ---
12
 
13
+ # Qwen1.5-4B-Chat
14
 
15
 
16
  ## Introduction
17
 
18
+ Qwen1.5 is the beta version of Qwen2, a transformer-based decoder-only language model pretrained on a large amount of data. In comparison with the previous released Qwen, the improvements include:
19
 
20
  * 6 model sizes, including 0.5B, 1.8B, 4B, 7B, 14B, and 72B;
21
  * Significant performance improvement in human preference for chat models;
 
27
  <br>
28
 
29
  ## Model Details
30
+ Qwen1.5 is a language model series including decoder language models of different model sizes. For each size, we release the base language model and the aligned chat model. It is based on the Transformer architecture with SwiGLU activation, attention QKV bias, group query attention, mixture of sliding window attention and full attention, etc. Additionally, we have an improved tokenizer adaptive to multiple natural languages and codes. For the beta version, temporarily we did not include GQA and the mixture of SWA and full attention.
31
 
32
  ## Training details
33
  We pretrained the models with a large amount of data, and we post-trained the models with both supervised finetuning and direct preference optimization. However, DPO leads to improvements in human preference evaluation but degradation in benchmark evaluation. In the very near future, we will fix both problems.
34
 
35
  ## Requirements
36
+ The code of Qwen1.5 has been in the latest Hugging face transformers and we advise you to install `transformers>=4.37.0`, or you might encounter the following error:
37
  ```
38
  KeyError: 'qwen2'
39
  ```
 
47
  device = "cuda" # the device to load the model onto
48
 
49
  model = AutoModelForCausalLM.from_pretrained(
50
+ "Qwen/Qwen1.5-4B-Chat",
51
  device_map="auto"
52
  )
53
+ tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen1.5-4B-Chat")
54
 
55
  prompt = "Give me a short introduction to large language model."
56
  messages = [
 
75
  response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
76
  ```
77
 
78
+ For quantized models, we advise you to use the GPTQ, AWQ, and GGUF correspondents, namely `Qwen1.5-4B-Chat-GPTQ`, `Qwen1.5-4B-Chat-AWQ`, and `Qwen1.5-4B-Chat-GGUF`.
79
 
80
 
81
  ## Limitations