Transformers
GGUF
llama
File size: 1,232 Bytes
775de30
 
 
 
 
 
 
ac172bb
84f1dc0
5e611ab
84f1dc0
a7da3c1
ac172bb
84f1dc0
a7da3c1
84f1dc0
ac172bb
3c96af7
ac172bb
84f1dc0
ac172bb
6ea3009
a7da3c1
6ea3009
a7da3c1
 
ac172bb
a7da3c1
ac172bb
a7da3c1
6ea3009
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
---
datasets:
  - PygmalionAI/PIPPA
  - ludis/geepeetee4
  - lemonilia/LimaRP
---

## GGUF

gguf quants for ludis/tsukasa-13b-qlora-limarp


## Prompting

https://rentry.org/tsukasa13b - reccomended prompts and gen settings

The current model version has been trained on prompts using three different roles, which are denoted by the following tokens: `<|system|>`, `<|user|>` and `<|model|>`.

The `<|system|>` prompt can be used to inject out-of-channel information behind the scenes, while the `<|user|>` prompt should be used to indicate user input. The `<|model|>` token should then be used to indicate that the model should generate a response. These tokens can happen multiple times and be chained up to form a conversation history.

## Training

base model (mistral-0.1-7b)

[axolotl](https://github.com/OpenAccess-AI-Collective/axolotl) was used for training
on a 4x nvidia a40 gpu cluster.

the a40 GPU cluster has been graciously provided by [Arc Compute](https://www.arccompute.io/).

rank 8 lora tune of mistralai/Mistral-7B-v0.1, first tuned on koishi commit 6e675d1 for one epoch then on limarp (without ponyville, lolicit, all the fallen, and eka's portal subsets) Version 2023-09-30 for 2 epochs in metharme format