Minueza-32Mx2-Chat

• Base model: Felladrin/Minueza-32M-Base
• Datasets used during SFT:
  • [ChatML] databricks/databricks-dolly-15k
  • [ChatML] euclaise/reddit-instruct-curated
  • [ChatML] THUDM/webglm-qa
  • [ChatML] starfishmedical/webGPT_x_dolly
  • [ChatML] LDJnr/Capybara
  • [ChatML] Open-Orca/SlimOrca-Dedup
  • [ChatML] HuggingFaceH4/ultrachat_200k
  • [ChatML] nvidia/HelpSteer
  • [ChatML] sablo/oasst2_curated
  • [ChatML] CohereForAI/aya_dataset
• Datasets used during DPO:
  • [ChatML] argilla/distilabel-capybara-dpo-7k-binarized
  • [ChatML] argilla/distilabel-intel-orca-dpo-pairs
  • [ChatML] argilla/ultrafeedback-binarized-preferences
  • [ChatML] sablo/oasst2_dpo_pairs_en
  • [ChatML] NeuralNovel/Neural-DPO
• License: Apache License 2.0

Recommended Prompt Format

<|im_start|>system
{system_message}<|im_end|>
<|im_start|>user
{user_message}<|im_end|>
<|im_start|>assistant

Recommended Inference Parameters

do_sample: true
temperature: 0.65
top_p: 0.55
top_k: 35
repetition_penalty: 1.176

Usage Example

from transformers import pipeline

generate = pipeline("text-generation", "Felladrin/Minueza-32Mx2-Chat")

messages = [
    {
        "role": "system",
        "content": "You are a helpful assistant who answers the user's questions with details and curiosity.",
    },
    {
        "role": "user",
        "content": "What are some potential applications for quantum computing?",
    },
]

prompt = generate.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)

output = generate(
    prompt,
    max_new_tokens=256,
    do_sample=True,
    temperature=0.65,
    top_k=35,
    top_p=0.55,
    repetition_penalty=1.176,
)

print(output[0]["generated_text"])

How it was trained

This model was trained with SFT Trainer and DPO Trainer, in several sessions, using the following settings:

For Supervised Fine-Tuning:

Hyperparameter	Value
Learning rate	2e-6
Total train batch size	16
Max. sequence length	2048
Weight decay	0.01
Warmup ratio	0.1
Optimizer	Adam with betas=(0.9,0.999) and epsilon=1e-08
Scheduler	cosine
Seed	42
Neftune Noise Alpha	5

For Direct Preference Optimization:

Hyperparameter	Value
Learning rate	5e-7
Total train batch size	16
Max. length	1024
Max. prompt length	512
Max. steps	200
Weight decay	0
Warmup ratio	0.1
Beta	0.1