Xin Liu commited on
Commit
7d325b4
·
1 Parent(s): 007486b

Signed-off-by: Xin Liu <sam@secondstate.io>

Files changed (1) hide show
  1. README.md +73 -9
README.md CHANGED
@@ -1,11 +1,75 @@
1
- # Yi-34B-Chat
 
 
 
 
 
 
 
 
 
 
 
2
 
3
- ## Prompt Template
 
 
 
 
 
 
4
 
5
- ```text
6
- <|im_start|>system
7
- {system_message}<|im_end|>
8
- <|im_start|>user
9
- {prompt}<|im_end|>
10
- <|im_start|>assistant
11
- ```
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: 01-ai/Yi-34B-Chat
3
+ inference: false
4
+ license: other
5
+ license_link: LICENSE
6
+ license_name: yi-license
7
+ model_creator: 01-ai
8
+ model_name: Yi 34B Chat
9
+ model_type: yi
10
+ pipeline_tag: text-generation
11
+ quantized_by: Second State Inc.
12
+ ---
13
 
14
+ <!-- header start -->
15
+ <!-- 200823 -->
16
+ <div style="width: auto; margin-left: auto; margin-right: auto">
17
+ <img src="https://github.com/second-state/LlamaEdge/raw/dev/assets/logo.svg" style="width: 100%; min-width: 400px; display: block; margin: auto;">
18
+ </div>
19
+ <hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
20
+ <!-- header end -->
21
 
22
+ # Yi-34B-Chat-GGUF
23
+
24
+ ## Original Model
25
+
26
+ [01-ai/Yi-34B-Chat](https://huggingface.co/01-ai/Yi-34B-Chat)
27
+
28
+ ## Run with LlamaEdge
29
+
30
+ - LlamaEdge version: [v0.2.8](https://github.com/second-state/LlamaEdge/releases/tag/0.2.8)
31
+
32
+ - Prompt template
33
+
34
+ - Prompt type: `chatml`
35
+
36
+ - Prompt string
37
+
38
+ ```text
39
+ <|im_start|>system
40
+ {system_message}<|im_end|>
41
+ <|im_start|>user
42
+ {prompt}<|im_end|>
43
+ <|im_start|>assistant
44
+ ```
45
+
46
+ - Reverse prompt: `<|im_end|>`
47
+
48
+ - Run as LlamaEdge service
49
+
50
+ ```bash
51
+ wasmedge --dir .:. --nn-preload default:GGML:AUTO:Yi-34B-Chat-ggml-model-q4_0.gguf llama-api-server.wasm -p chatml -r '<|im_end|>'
52
+ ```
53
+
54
+ - Run as LlamaEdge command app
55
+
56
+ ```bash
57
+ wasmedge --dir .:. --nn-preload default:GGML:AUTO:Yi-34B-Chat-ggml-model-q4_0.gguf llama-chat.wasm -p chatml -r '<|im_end|>'
58
+ ```
59
+
60
+ ## Quantized GGUF Models
61
+
62
+ | Name | Quant method | Bits | Size | Use case |
63
+ | ---- | ---- | ---- | ---- | ----- |
64
+ | [Yi-34B-Chat-Q2_K.gguf](https://huggingface.co/second-state/Yi-34B-Chat-GGUF/blob/main/Yi-34B-Chat-Q2_K.gguf) | Q2_K | 2 | 12.8 GB| smallest, significant quality loss - not recommended for most purposes |
65
+ | [Yi-34B-Chat-Q3_K_L.gguf](https://huggingface.co/second-state/Yi-34B-Chat-GGUF/blob/main/Yi-34B-Chat-Q3_K_L.gguf) | Q3_K_L | 3 | 18.1 GB| small, substantial quality loss |
66
+ | [Yi-34B-Chat-Q3_K_M.gguf](https://huggingface.co/second-state/Yi-34B-Chat-GGUF/blob/main/Yi-34B-Chat-Q3_K_M.gguf) | Q3_K_M | 3 | 16.7 GB| very small, high quality loss |
67
+ | [Yi-34B-Chat-Q3_K_S.gguf](https://huggingface.co/second-state/Yi-34B-Chat-GGUF/blob/main/Yi-34B-Chat-Q3_K_S.gguf) | Q3_K_S | 3 | 15.0 GB| very small, high quality loss |
68
+ | [Yi-34B-Chat-Q4_0.gguf](https://huggingface.co/second-state/Yi-34B-Chat-GGUF/blob/main/Yi-34B-Chat-Q4_0.gguf) | Q4_0 | 4 | 19.5 GB| legacy; small, very high quality loss - prefer using Q3_K_M |
69
+ | [Yi-34B-Chat-Q4_K_M.gguf](https://huggingface.co/second-state/Yi-34B-Chat-GGUF/blob/main/Yi-34B-Chat-Q4_K_M.gguf) | Q4_K_M | 4 | 20.7 GB| medium, balanced quality - recommended |
70
+ | [Yi-34B-Chat-Q4_K_S.gguf](https://huggingface.co/second-state/Yi-34B-Chat-GGUF/blob/main/Yi-34B-Chat-Q4_K_S.gguf) | Q4_K_S | 4 | 19.6 GB| small, greater quality loss |
71
+ | [Yi-34B-Chat-Q5_0.gguf](https://huggingface.co/second-state/Yi-34B-Chat-GGUF/blob/main/Yi-34B-Chat-Q5_0.gguf) | Q5_0 | 5 | 23.7 GB| legacy; medium, balanced quality - prefer using Q4_K_M |
72
+ | [Yi-34B-Chat-Q5_K_M.gguf](https://huggingface.co/second-state/Yi-34B-Chat-GGUF/blob/main/Yi-34B-Chat-Q5_K_M.gguf) | Q5_K_M | 5 | 24.3 GB| large, very low quality loss - recommended |
73
+ | [Yi-34B-Chat-Q5_K_S.gguf](https://huggingface.co/second-state/Yi-34B-Chat-GGUF/blob/main/Yi-34B-Chat-Q5_K_S.gguf) | Q5_K_S | 5 | 23.7 GB| large, low quality loss - recommended |
74
+ | [Yi-34B-Chat-Q6_K.gguf](https://huggingface.co/second-state/Yi-34B-Chat-GGUF/blob/main/Yi-34B-Chat-Q6_K.gguf) | Q6_K | 6 | 28.2 GB| very large, extremely low quality loss |
75
+ | [Yi-34B-Chat-Q8_0.gguf](https://huggingface.co/second-state/Yi-34B-Chat-GGUF/blob/main/Yi-34B-Chat-Q8_0.gguf) | Q8_0 | 8 | 36.5 GB| very large, extremely low quality loss - not recommended |