File size: 1,300 Bytes
d6031e6
 
 
 
 
 
 
 
 
 
b4082e3
d6031e6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
license: llama3
base_model: OwenArli/ArliAI-Llama-3-8B-Cumulus-v0.3
pipeline_tag: text-generation
tags:
- llama
- conversational
---

# QuantFactory/ArliAI-Llama-3-8B-Cumulus-v0.3-GGUF
This is quantized version of [OwenArli/ArliAI-Llama-3-8B-Cumulus-v0.3](https://huggingface.co./OwenArli/ArliAI-Llama-3-8B-Cumulus-v0.3) created using llama.cpp


# Model Description
Based on Meta-Llama-3-8b-Instruct, and is governed by Meta Llama 3 License agreement:
https://huggingface.co./meta-llama/Meta-Llama-3-8B-Instruct


In terms of reasoning and intelligence, this model is probably a bit worse than the OG model because of the decensoring. However, this model is better at long back and forth chats and will refuse less.


OpenLLM Benchmark:


Training:
- Full 8192 sequence length.
- Training duration is around 2 days on an RTX 4090, using 4-bit loading and Qlora 64-rank 64-alpha resulting in ~2% trainable weights.


Instruct format:
```
<|begin_of_text|><|start_header_id|>system<|end_header_id|>

{{ system_prompt }}<|eot_id|><|start_header_id|>user<|end_header_id|>

{{ user_message_1 }}<|eot_id|><|start_header_id|>assistant<|end_header_id|>

{{ model_answer_1 }}<|eot_id|><|start_header_id|>user<|end_header_id|>

{{ user_message_2 }}<|eot_id|><|start_header_id|>assistant<|end_header_id|>
```