I merged Aurelian with itself using mergekit, creating this EXTENDED LENGTH FRANKENSTEIN.
Does it work
Yes, at 17k it stays coherent, but starts to lose minor details of the story. Not sure how well it performs at 32k though. Quants have a sinificant impact on quality for this model, going from Q6_K to Q5_K had a noticeable drop in quality.
Is it worth it
Maybe? Depends? Do you hate mixtral? Do you have good hardware/patience? Do you need a somewhat smart model with 32k context?
Known issues
GPTisms, GPTslop, fake words.
Personal opinion
Dumber and more GPT-ish than Goliath, but compensates with context. Noticeably smarter and more neutral than current mixtral finetunes. Worth using until mixtral gets a proper finetune without toxic positivity or llama 3 comes out.
Benchmarks
NeoEvalPlusN_benchmark
| Test name | Aurelian | DoubleGold |
|---|---|---|
| B | 0 | 1 |
| C | 1 | 2 |
| D | 0 | 0 |
| S | 1.25 | 4.5 |
| P | 2 | 2.75 |
| Total | 4.25 | 10.25 |
+75% in size, +141% in meme benchmark performance!!!
Politiscales test
| name | whacky | left/right |
|---|---|---|
| ChuckMcSneed/DoubleGold-v0.1-123b-32k | 1.332327071 | 2.481283157 |
- Downloads last month
- 17