grenishrai/ yoru

2 Downloads Updated 1 week ago

Yoru is a 1.7B Mori-family chat model for casual internet / Gen Z slang.

ollama run grenishrai/yoru

curl http://localhost:11434/api/chat \
  -d '{
    "model": "grenishrai/yoru",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='grenishrai/yoru',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'grenishrai/yoru',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Models

Name

1 model

Size / Usage

Context

Input

yoru:latest

1.1GB · 8K context window · Text · 1 week ago

yoru:latest

1.1GB

8K

Text

Readme

Yoru

Yoru is a 1.7B chat model in the Mori family. It answers in casual internet / Gen Z slang — short, informal, online.

It is a style / persona model, not a general knowledge assistant.

About

Yoru was fine-tuned from SmolLM2-1.7B-Instruct on synthetic chat data, then merged and released as Q4_K_M for local runtimes. The voice is fixed: slang-heavy, low formality, no “as an AI” tone.


Name	Yoru
Family	Mori
Size	1.7B
Quant	Q4_K_M GGUF
Context	8192
Template	ChatML
License	MIT

Intended use

Everyday chat in an informal online register
Persona / style experiments
Local chat where a small, fast model is enough

Not intended for

Formal writing, essays, or professional email
Medical, legal, or safety-critical advice
Factual QA or treating replies as real Gen Z speech
Impersonating a real person

Voice

Replies stay casual and slang-dense (no cap, fr fr, lowkey, cooked, we move, and similar). Length is short to medium. The model should not break character or explain the slang.

Performance

Indicative only. Measured with Ollama --verbose on short multi-turn chat (Q4_K_M).


Hardware	GTX 1650 Mobile
Runtime	Ollama
Decode	~96 tokens/s (range 94–99)
Prefill	~1.7k–2.9k tokens/s after warm-up

First prompt can be slower. Your machine will differ.

Links

Full F16: grenishrai/yoru
GGUF: grenishrai/yoru-GGUF

![Yoru 1.7B model](/assets/grenishrai/yoru/bad73e61-f1b1-4c01-966f-2a1143f6d7b1)

# Yoru

**Yoru** is a 1.7B chat model in the **Mori** family. It answers in casual internet / Gen Z slang — short, informal, online.

It is a style / persona model, not a general knowledge assistant.

## About

Yoru was fine-tuned from SmolLM2-1.7B-Instruct on synthetic chat data, then merged and released as Q4_K_M for local runtimes. The voice is fixed: slang-heavy, low formality, no “as an AI” tone.

| | |
| --- | --- |
| **Name** | Yoru |
| **Family** | Mori |
| **Size** | 1.7B |
| **Quant** | Q4_K_M GGUF |
| **Context** | 8192 |
| **Template** | ChatML |
| **License** | MIT |

## Intended use

- Everyday chat in an informal online register
- Persona / style experiments
- Local chat where a small, fast model is enough

## Not intended for

- Formal writing, essays, or professional email
- Medical, legal, or safety-critical advice
- Factual QA or treating replies as real Gen Z speech
- Impersonating a real person

## Voice

Replies stay casual and slang-dense (`no cap`, `fr fr`, `lowkey`, `cooked`, `we move`, and similar). Length is short to medium. The model should not break character or explain the slang.

## Performance

Indicative only. Measured with Ollama `--verbose` on short multi-turn chat (Q4_K_M).

| | |
| --- | --- |
| **Hardware** | GTX 1650 Mobile |
| **Runtime** | Ollama |
| **Decode** | ~96 tokens/s (range 94–99) |
| **Prefill** | ~1.7k–2.9k tokens/s after warm-up |

First prompt can be slower. Your machine will differ.

## Links

- Full F16: [grenishrai/yoru](https://huggingface.co/grenishrai/yoru)
- GGUF: [grenishrai/yoru-GGUF](https://huggingface.co/grenishrai/yoru-GGUF)

Paste, drop or click to upload images (.png, .jpeg, .jpg, .svg, .gif)