Update README.md
Browse files
README.md
CHANGED
|
@@ -1,12 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
[axolotl](https://github.com/OpenAccess-AI-Collective/axolotl) was used for training
|
| 2 |
on a 2x nvidia a40 gpu cluster.
|
| 3 |
|
| 4 |
the a40 GPU cluster has been graciously provided by [Arc Compute](https://www.arccompute.io/).
|
| 5 |
-
rank 8 lora merge
|
| 6 |
|
| 7 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
|
| 9 |
-
this is the format
|
| 10 |
|
| 11 |
```
|
| 12 |
### System:
|
|
@@ -14,11 +30,4 @@ this is the format
|
|
| 14 |
### Instruction:
|
| 15 |
|
| 16 |
### Response:
|
| 17 |
-
```
|
| 18 |
-
|
| 19 |
-
trained on subset of koishi
|
| 20 |
-
|
| 21 |
-
- asss
|
| 22 |
-
- dolly
|
| 23 |
-
- hh-rlhf
|
| 24 |
-
- wizard evol
|
|
|
|
| 1 |
+
---
|
| 2 |
+
datasets:
|
| 3 |
+
- ewof/koishi-instruct-metharme
|
| 4 |
+
---
|
| 5 |
+
|
| 6 |
+
## Training
|
| 7 |
+
|
| 8 |
[axolotl](https://github.com/OpenAccess-AI-Collective/axolotl) was used for training
|
| 9 |
on a 2x nvidia a40 gpu cluster.
|
| 10 |
|
| 11 |
the a40 GPU cluster has been graciously provided by [Arc Compute](https://www.arccompute.io/).
|
|
|
|
| 12 |
|
| 13 |
+
trained on subset of koishi
|
| 14 |
+
|
| 15 |
+
- asss
|
| 16 |
+
- dolly
|
| 17 |
+
- hh-rlhf
|
| 18 |
+
- wizard evol
|
| 19 |
+
|
| 20 |
+
## Base Model
|
| 21 |
+
|
| 22 |
+
rank 8 lora tune of mistralai/Mistral-7B-v0.1
|
| 23 |
+
|
| 24 |
+
## Prompting
|
| 25 |
|
|
|
|
| 26 |
|
| 27 |
```
|
| 28 |
### System:
|
|
|
|
| 30 |
### Instruction:
|
| 31 |
|
| 32 |
### Response:
|
| 33 |
+
```
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|