/lmg/ Official Best Models To Use Guide
This is a collaborative project that anyone can edit. You WILL contribute, and help keep this project up-to-date for the benefit of the whole general! The document is read-only, explanation how to contribute below!
How to contribute: Click edit in the bottom left corner, copy the whole thing (important to keep the markdown intact, copying the preview that is seen by default would break the document), make a new rentry with your edits (do not share the edit code, or edit the rentry, make new ones instead!), and post near the end of the thread for OP to put into the new one.
Current table is shamelessly ripped from the pygmalion wiki which used to be in the OP and frequently updated, but since has been removed and isn't updated frequently anymore, feel free to remove this line when you find better models to add!
| Parameter size | Model | GGUF huggingface link | exl2 huggingface link |
|---|---|---|---|
| 7B | Una-TheBeagle-7B-v1 | GGUF | exl2 |
| 8B | Hathor_Respawn-L3-8B-v0.8 | GGUF | exl2 |
| 11B | Fimbulvetr-11B-v2 | GGUF | exl2 |
| 12B | Rocinante 12B v1.1 | GGUF | exl2 |
| 20B | Rose | GGUF | exl2 |
| 20B | psyonic-cetacean | GGUF | exl2 |
| 35B | c4ai-commanr-r-v01 | GGUF | exl2 |
| 70B | Midnight Miqu | GGUF | IDK if the exl2 quants on HF are any good so I didn't put them here |
| 103B | c4ai-command-r-plus | GGUF | exl2 |
| 120B | Goliath | GGUF | exl2 |