RichardErkhov/flammenai_-_Mahou-1.3-gemma2-9B-gguf

Quantization made by Richard Erkhov.

Github

Discord

Request more models

Mahou-1.3-gemma2-9B - GGUF

• Model creator: https://huggingface.co/flammenai/
• Original model: https://huggingface.co/flammenai/Mahou-1.3-gemma2-9B/

Name	Quant method	Size
Mahou-1.3-gemma2-9B.Q2_K.gguf	Q2_K	3.54GB
Mahou-1.3-gemma2-9B.IQ3_XS.gguf	IQ3_XS	3.86GB
Mahou-1.3-gemma2-9B.IQ3_S.gguf	IQ3_S	4.04GB
Mahou-1.3-gemma2-9B.Q3_K_S.gguf	Q3_K_S	4.04GB
Mahou-1.3-gemma2-9B.IQ3_M.gguf	IQ3_M	4.19GB
Mahou-1.3-gemma2-9B.Q3_K.gguf	Q3_K	4.43GB
Mahou-1.3-gemma2-9B.Q3_K_M.gguf	Q3_K_M	4.43GB
Mahou-1.3-gemma2-9B.Q3_K_L.gguf	Q3_K_L	4.78GB
Mahou-1.3-gemma2-9B.IQ4_XS.gguf	IQ4_XS	4.86GB
Mahou-1.3-gemma2-9B.Q4_0.gguf	Q4_0	5.07GB
Mahou-1.3-gemma2-9B.IQ4_NL.gguf	IQ4_NL	5.1GB
Mahou-1.3-gemma2-9B.Q4_K_S.gguf	Q4_K_S	5.1GB
Mahou-1.3-gemma2-9B.Q4_K.gguf	Q4_K	5.37GB
Mahou-1.3-gemma2-9B.Q4_K_M.gguf	Q4_K_M	5.37GB
Mahou-1.3-gemma2-9B.Q4_1.gguf	Q4_1	5.55GB
Mahou-1.3-gemma2-9B.Q5_0.gguf	Q5_0	6.04GB
Mahou-1.3-gemma2-9B.Q5_K_S.gguf	Q5_K_S	6.04GB
Mahou-1.3-gemma2-9B.Q5_K.gguf	Q5_K	6.19GB
Mahou-1.3-gemma2-9B.Q5_K_M.gguf	Q5_K_M	6.19GB
Mahou-1.3-gemma2-9B.Q5_1.gguf	Q5_1	6.52GB
Mahou-1.3-gemma2-9B.Q6_K.gguf	Q6_K	7.07GB
Mahou-1.3-gemma2-9B.Q8_0.gguf	Q8_0	9.15GB

Original model description:

library_name: transformers license: gemma base_model: - google/gemma-2-9b datasets: - flammenai/FlameMix-DPO-v1 - flammenai/MahouMix-v1 - flammenai/Grill-Flammen-v1_chatML

Mahou-1.3-gemma2-9B

Mahou is designed to provide short messages in a conversational context. It is capable of casual conversation and character roleplay.

Chat Format

This model has been trained to use ChatML format. Note the additional tokens in tokenizer_config.json.

<|im_start|>system
{{system}}<|im_end|>
<|im_start|>{{char}}
{{message}}<|im_end|>
<|im_start|>{{user}}
{{message}}<|im_end|>

Roleplay Format

• Speech without quotes.
• Actions in *asterisks*

*leans against wall cooly* so like, i just casted a super strong spell at magician academy today, not gonna lie, felt badass.

SillyTavern Settings

1. Use ChatML for the Context Template.
2. Enable Instruct Mode.
3. Use the Mahou preset.
4. Recommended Additonal stopping strings: ["\n", "<|", "</"]

Method

Finetuned for 3 epochs using an A100 on Google Colab.

Fine-tune Llama 3 with ORPO - Maxime Labonne