Hastagaras
commited on
Commit
•
6f0c649
1
Parent(s):
9963e2e
Update README.md
Browse files
README.md
CHANGED
@@ -23,11 +23,11 @@ And thanks for all the feedback from the previous model, it helps a lot. (I coul
|
|
23 |
|
24 |
**More Details:**
|
25 |
|
26 |
-
*
|
27 |
-
* Next
|
28 |
-
*
|
29 |
-
* Next,
|
30 |
-
* And then
|
31 |
|
32 |
And that's it.
|
33 |
|
|
|
23 |
|
24 |
**More Details:**
|
25 |
|
26 |
+
* This model is based on [this model](https://huggingface.co/Hastagaras/Jamet-8B-L3-MK.II-LV), which is based on the UltimateAnjir model. It shares the same creative, cheerful, and positive tendencies. Then I merged it with Llama 3 Instruct.
|
27 |
+
* Next is DPO! to reduce the cheerfulness, emojis, and positivity. (This is based on the [Jamet MK.II Feedback](https://huggingface.co/Hastagaras/Jamet-8B-L3-MK.II) regarding positivity.) I trained a QLora with about 1,000 prompts from Alpaca to generate a dataset, selecting the ones with emojis, removing the emojis using regex and moving those without emojis to the chosen, and the responses with emojis to the rejected.(With unsloth)
|
28 |
+
* Then, I applied the [Abomination Lora from Blackroot](https://huggingface.co/Blackroot/Llama-3-8B-Abomination-LORA).
|
29 |
+
* Next, I applied the [Anjir Adapter](https://huggingface.co/Hastagaras/ANJIR-ADAPTER-64) (64 Rank version with reduced Alpha) to improve formatting while retaining the previous Lora influences. (This is based on the Anjir Feedback, which suggests that Anjir has better formatting than the Halu Blackroot.)
|
30 |
+
* And then merged the model with the Anjrit model. (I won't release the Anjrit model as it struggles with longer contexts. I'm only interested in its no refusals storytelling abilities, but you can find a brief overview of the model on my [Anjir model](https://huggingface.co/Hastagaras/Anjir-8B-L3) page.)
|
31 |
|
32 |
And that's it.
|
33 |
|