README.md

---
license: llama3
library_name: transformers
tags:
- not-for-all-audiences
---
<div align="left">
  <img src="https://huggingface.co/Hastagaras/Jamet-8B-L3-MK.V-Blackroot/resolve/main/MARK V BR.png" width="850"/>
</div>

### JAMETSSS

[GGUF](https://huggingface.co/backyardai/Jamet-8B-L3-MK.V-Blackroot-GGUF) made available by [Backyard AI](https://huggingface.co/backyardai)

[ANOTHER GGUF](https://huggingface.co/mradermacher/Jamet-8B-L3-MK.V-Blackroot-GGUF) made available by [mradermacher](https://huggingface.co/mradermacher)


Different base model, different methods (without model stock merge, because it's not good...at least for me, it gives repeating sentences or words at 0 temperature, that also happened with the Halu Blackroot). This model maybe more similar to the Anjir model.

This model has been on a long journey; I have like...9 variations of it. I tested all of them at Q4_K_M and decided to release this one, this is variation number 7.

And thanks for all the feedback from the previous model, it helps a lot. (I couldn't fix the issue with German because I don't know how to fix it since I can't speak German, heck, even my English is bad.)

**More Details:**

* Based on [this model](https://huggingface.co/Hastagaras/Jamet-8B-L3-MK.II-LV), and merge that with Llama 3 Instruct.
* Next step is DPO! to reduce the cheerfulness and emoji, using around 1k prompts from alpaca to generate a dataset with the model, 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)
* And then apply the [Abomination Lora from Blackroot](https://huggingface.co/Blackroot/Llama-3-8B-Abomination-LORA).
* Next, apply [Anjir Adapter](https://huggingface.co/Hastagaras/ANJIR-ADAPTER-64) (64 Rank version with reduced Alpha to help with formatting while retaining the previous lora influence).
* And then merge again with the Anjrit model (I don't release the model since it's very bad at longer context, I just want the storytelling ability but you can read small info of the model at my [Anjir model](https://huggingface.co/Hastagaras/Anjir-8B-L3)).

And that's it.

**Notes:**
* I'm not responsible for anything.
* This is an RP and Storytelling model.
* You can write your feedback at the discussion, so i can improve my models.
* Like all of my previous models, higher temperatures will be incoherent, so use around 0.85-1.05.(I've been trying to fix this since Halu Blackroot but haven't had much luck, I think merging the base with Llama 3 Instruct helped a lot.)

<div align="left">
  <img src="https://huggingface.co/Hastagaras/Jamet-8B-L3-MK.V-Blackroot/resolve/main/chrome_UcaBMEjkH3.png" width="850"/>
</div>

<div align="left">
  <img src="https://huggingface.co/Hastagaras/Jamet-8B-L3-MK.V-Blackroot/resolve/main/chrome_BH9VUDnqZy.png" width="850"/>
</div>
^Sometimes it still gives a response with emojis (4Bit).