File size: 2,135 Bytes
b7fcf5d
 
debadec
ef9950d
b7fcf5d
538aa5a
ef9950d
362713e
538aa5a
ef9950d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
362713e
ef9950d
 
362713e
 
 
ef9950d
 
7cf8a10
 
 
 
 
 
 
 
ef9950d
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
## This is an experimental example of an .aiml file. 
# Everything you see hereis a work in progress, dont use this in production. Please.

{
  "aiml": {
    "version": "0.1.1",
    "last_modified": "September 9th 2023, 01:30",
    "properties": {
      "purpose": "text-generation-llm",
      "custom_key": "custom_value"
    }
  },
  "modelProfile": {
    "model": {
      "name": "jphme/llama-2-13b-chat-german",
      "purpose": "text-generation",
      "created_by": "hf.co/jphme",
      "category": "llm",
      "class": "llama2",
      "architecture": "transformers",
      "finetuned": "yes",
      "finetuned_type": "chat, instruct",
      "[category]_metadata": {    # e.g. "llm_metadata", specific metadata for model type
        "parameters": "70b",
        "architecture": "transformers",
        "fileFormats": "gguf, ggmlv3, gptq",
        "custom_key": "custom_value"
      }
    },
    "finetuning": {
      "method": "full",
      "datasets": [
        "Prorietary german conversation dataset - [View Dataset](https://huggingface.co./datasets)"",
        "German Squad - [View Dataset](https://huggingface.co./datasets)",
        "German Legal Squad - [View Dataset](https://huggingface.co./datasets)"",
      ],
      "description": "According to the model creator, the dataset was finetuned with several instruct-based datasets in German language and augmented with false RAG data to improve model factuality.""
    },
    "quantization": {
      "types": ["8 Bit", "5 Bit (k_m)"],
      "formats": ["GGUF"],
      "tools_used": "llama.cpp (commit 9e20231) - for quantization to 8, 5 and 4 bit"
    },
    "deployment": {
      "id": "1",
      "title": "Bare Metal - Only CPU (64 Bit Intel/AMD/ARM)",
      "description": "Use run.sh to deploy the model. Note: The run.sh script is not yet published in this repo. It will be made available within the next few days."
      "cpu_architecture": "x86-64, arm64",
      "os": "Mac OS, Linux (Ubuntu), Linux (Debian), Windows",
      "install": "chmod +x install.sh && ./install.sh",
      "start": "./run.sh",
      "stop": "systemctl stop run_aiml.service",
    }
  }
}