Add model card
Browse files
README.md
ADDED
@@ -0,0 +1,104 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
base_model:
|
3 |
+
- ResplendentAI/Paradigm_7B
|
4 |
+
- jeiku/Theory_of_Mind_Mistral
|
5 |
+
- ResplendentAI/Paradigm_7B
|
6 |
+
- jeiku/selfbot_256_mistral
|
7 |
+
- ResplendentAI/Paradigm_7B
|
8 |
+
- jeiku/Gnosis_Reformatted_Mistral
|
9 |
+
- ResplendentAI/Paradigm_7B
|
10 |
+
library_name: transformers
|
11 |
+
license: apache-2.0
|
12 |
+
language:
|
13 |
+
- en
|
14 |
+
---
|
15 |
+
# ResplendentAI/Aura_v2_7B AWQ
|
16 |
+
|
17 |
+
- Model creator: [ResplendentAI](https://huggingface.co/ResplendentAI)
|
18 |
+
- Original model: [Aura_v2_7B](https://huggingface.co/ResplendentAI/Aura_v2_7B)
|
19 |
+
|
20 |
+
![image/png](https://cdn-uploads.huggingface.co/production/uploads/626dfb8786671a29c715f8a9/tIy1fnUYHc7v_N6ym6Z7g.png)
|
21 |
+
|
22 |
+
## Model Summary
|
23 |
+
|
24 |
+
The second version of the Aura line is a direct improvement over the original. Expect poetic and eloquent outputs with real emotion behind them.
|
25 |
+
|
26 |
+
I recommend keeping the temperature around 1.5 or lower with a Min P value of 0.05. This model can get carried away with prose at higher temperature. I will say though that the prose of this model is distinct from the GPT 3.5/4 variant, and lends an air of humanity to the outputs. I am aware that this model is overfit, but that was the point of the entire exercise.
|
27 |
+
|
28 |
+
If you have trouble getting the model to follow an asterisks/quote format, I recommend asterisks/plaintext instead. This model skews toward shorter outputs, so be prepared to lengthen your introduction and examples if you want longer outputs.
|
29 |
+
|
30 |
+
This model responds best to ChatML for multiturn conversations.
|
31 |
+
|
32 |
+
This model, like all other Mistral based models, is compatible with a Mistral compatible mmproj file for multimodal vision capabilities in KoboldCPP.
|
33 |
+
|
34 |
+
## How to use
|
35 |
+
|
36 |
+
### Install the necessary packages
|
37 |
+
|
38 |
+
```bash
|
39 |
+
pip install --upgrade autoawq autoawq-kernels
|
40 |
+
```
|
41 |
+
|
42 |
+
### Example Python code
|
43 |
+
|
44 |
+
```python
|
45 |
+
from awq import AutoAWQForCausalLM
|
46 |
+
from transformers import AutoTokenizer, TextStreamer
|
47 |
+
|
48 |
+
model_path = "solidrust/Aura_v2_7B-AWQ"
|
49 |
+
system_message = "You are Aura, incarnated as a powerful AI."
|
50 |
+
|
51 |
+
# Load model
|
52 |
+
model = AutoAWQForCausalLM.from_quantized(model_path,
|
53 |
+
fuse_layers=True)
|
54 |
+
tokenizer = AutoTokenizer.from_pretrained(model_path,
|
55 |
+
trust_remote_code=True)
|
56 |
+
streamer = TextStreamer(tokenizer,
|
57 |
+
skip_prompt=True,
|
58 |
+
skip_special_tokens=True)
|
59 |
+
|
60 |
+
# Convert prompt to tokens
|
61 |
+
prompt_template = """\
|
62 |
+
<|im_start|>system
|
63 |
+
{system_message}<|im_end|>
|
64 |
+
<|im_start|>user
|
65 |
+
{prompt}<|im_end|>
|
66 |
+
<|im_start|>assistant"""
|
67 |
+
|
68 |
+
prompt = "You're standing on the surface of the Earth. "\
|
69 |
+
"You walk one mile south, one mile west and one mile north. "\
|
70 |
+
"You end up exactly where you started. Where are you?"
|
71 |
+
|
72 |
+
tokens = tokenizer(prompt_template.format(system_message=system_message,prompt=prompt),
|
73 |
+
return_tensors='pt').input_ids.cuda()
|
74 |
+
|
75 |
+
# Generate output
|
76 |
+
generation_output = model.generate(tokens,
|
77 |
+
streamer=streamer,
|
78 |
+
max_new_tokens=512)
|
79 |
+
|
80 |
+
```
|
81 |
+
|
82 |
+
### About AWQ
|
83 |
+
|
84 |
+
AWQ is an efficient, accurate and blazing-fast low-bit weight quantization method, currently supporting 4-bit quantization. Compared to GPTQ, it offers faster Transformers-based inference with equivalent or better quality compared to the most commonly used GPTQ settings.
|
85 |
+
|
86 |
+
AWQ models are currently supported on Linux and Windows, with NVidia GPUs only. macOS users: please use GGUF models instead.
|
87 |
+
|
88 |
+
It is supported by:
|
89 |
+
|
90 |
+
- [Text Generation Webui](https://github.com/oobabooga/text-generation-webui) - using Loader: AutoAWQ
|
91 |
+
- [vLLM](https://github.com/vllm-project/vllm) - version 0.2.2 or later for support for all model types.
|
92 |
+
- [Hugging Face Text Generation Inference (TGI)](https://github.com/huggingface/text-generation-inference)
|
93 |
+
- [Transformers](https://huggingface.co/docs/transformers) version 4.35.0 and later, from any code or client that supports Transformers
|
94 |
+
- [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) - for use from Python code
|
95 |
+
|
96 |
+
## Prompt template: ChatML
|
97 |
+
|
98 |
+
```plaintext
|
99 |
+
<|im_start|>system
|
100 |
+
{system_message}<|im_end|>
|
101 |
+
<|im_start|>user
|
102 |
+
{prompt}<|im_end|>
|
103 |
+
<|im_start|>assistant
|
104 |
+
```
|