Summary

This is a Gemma 2 27B qlora trained using a slightly modified version of the VNTL-v3.1-1k dataset, concatenated with the VNTL-Chat dataset.

This was trained mostly with the same hyperparameters as the VNTL LLaMA 3 8B qlora, the differences are:

  • Changed batch size and gradient accumulation from 6/5 to 15/2, this was done mostly for training efficiency since this qlora was trained using an A100.

As the 8B qlora, this version also includes the "chat mode" for Japanese grammar questions.

Translation Prompt

This is a prompt example for translation:

<<METADATA>>
[character] Name: Uryuu Sakuno (็“œ็”Ÿ ๆกœไนƒ) | Gender: Female
[character] Name: Uryuu Shingo (็“œ็”Ÿ ๆ–ฐๅพ) | Gender: Male | Aliases: Onii-chan (ใŠๅ…„ใกใ‚ƒใ‚“)
<<TRANSLATE>>
<<JAPANESE>>
่ฒทใ„็‰ฉใซ่กŒใฃใŸใฃใใ‚Šใชใ‹ใชใ‹ๅธฐใฃใฆใ“ใชใ„ใจๆ€ใฃใŸใ‚‰ใ€ใ‚„ใฃใฑใ‚Šใ€่ฟทๅญใซใชใฃใฆใ„ใŸใ‚‰ใ—ใ„ใ€‚
<<ENGLISH>>
I thought she was taking too long to come back from shopping, and it turns out she had gotten lost as I expected.<eos>
<<JAPANESE>>
[ๆ–ฐๅพ]: ใ€Œใใฃใ‹ใ€‚ใ†ใ‚“ใ€้›ป่ฉฑใ—ใฆใฟใฆใ‚ˆใ‹ใฃใŸใ€
<<ENGLISH>>
[Shingo]: "Is that so? Well, I'm glad I called you then."<eos>
<<JAPANESE>>
ๅฝผๅฅณใฏๅŸบๆœฌ็š„ใซ่ฝใก็€ใ„ใŸ้ ญใฎใ„ใ„ๅญใชใ‚“ใ ใ‘ใฉใ€็ตๆง‹ใช็ญ‹้‡‘ใฎๅ…ฅใฃใŸๆ–นๅ‘้Ÿณ็—ดใงใ‚‚ใ‚ใ‚‹ใ€‚
<<ENGLISH>>
She's basically a smart and level-headed girl, but she also has a very bad sense of direction.<eos>
<<JAPANESE>>
้ ปๅบฆใ“ใใ•ใปใฉใงใฏใชใ„ใ‚‚ใฎใฎใ€ใฒใจใŸใณ่ฟทๅญใซใชใ‚‹ใจใ€ใจใซใ‹ใๆดพๆ‰‹ใซ้“ใซ่ฟทใฃใฆใ—ใพใ†ใ€‚
<<ENGLISH>>
Though not very frequently, once she gets lost, she gets lost in a spectacular way.<eos>
<<JAPANESE>>
[ๆกœไนƒ]: ใ€Žโ€ฆโ€ฆใ”ใ‚ใ‚“ใ€
<<ENGLISH>>
[Sakuno]: "... Sorry."<eos>
<<JAPANESE>>
[ๆ–ฐๅพ]: ใ€Œใ†ใ†ใ‚“ใ€ใ“ใ†่จ€ใฃใกใ‚ƒใชใ‚“ใ ใ‘ใฉใ€่ฟทๅญใงใ‚ˆใ‹ใฃใŸใ‚ˆใ€‚ ๆกœไนƒใฏๅฏๆ„›ใ„ใ‹ใ‚‰ใ€ใ„ใ‚ใ„ใ‚ๅฟƒ้…ใ—ใกใ‚ƒใฃใฆใŸใ‚“ใ ใžไฟบใ€
<<ENGLISH>>

The generated translation for that prompt, with temperature 0, is:

[Shingo]: "No, it's fine. I'm glad you got lost, actually. You're so cute, I was worried about you."

Chat Prompt

This is a prompt example for chat:

<<CHAT>>
<<HUMAN>>
Hi! Can you break down this sentence for me? Like, what is the meaning of "ๆกœไนƒใฏๅฏๆ„›ใ„ใ‹ใ‚‰"?
"ใ†ใ†ใ‚“ใ€ใ“ใ†่จ€ใฃใกใ‚ƒใชใ‚“ใ ใ‘ใฉใ€่ฟทๅญใงใ‚ˆใ‹ใฃใŸใ‚ˆใ€‚ๆกœไนƒใฏๅฏๆ„›ใ„ใ‹ใ‚‰ใ€ใ„ใ‚ใ„ใ‚ๅฟƒ้…ใ—ใกใ‚ƒใฃใฆใŸใ‚“ใ ใžไฟบใ€"
<<LLM>>

The generated response for that prompt, with temperature 0, is:

Certainly! Let's start with your first question. "ๆกœไนƒใฏๅฏๆ„›ใ„ใ‹ใ‚‰" is a phrase that consists of three parts:
1. "ๆกœไนƒ" is a proper noun, likely a person's name.
2. "ใฏ" is a particle that marks the topic of the sentence, in this case, "ๆกœไนƒ."
3. "ๅฏๆ„›ใ„" is an adjective that means "cute" or "adorable."
4. "ใ‹ใ‚‰" is a conjunction that indicates a reason or cause.
So, the literal translation of "ๆกœไนƒใฏๅฏๆ„›ใ„ใ‹ใ‚‰" would be "Because Sakura is cute." It suggests that the speaker is about to provide a reason or explanation based on the fact that the person named Sakura is cute.
Downloads last month
4
Inference Examples
Inference API (serverless) does not yet support peft models for this pipeline type.

Model tree for lmg-anon/vntl-gemma2-27b-qlora

Base model

google/gemma-2-27b
Adapter
(7)
this model

Datasets used to train lmg-anon/vntl-gemma2-27b-qlora