sujet-ai/Sujet-Finance-8B-v0.1

Hugging Face
TEXT GENERATIONPricing:Input $0.37 / Cached $0.074 / Output $0.38Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:8kTool Calling:SupportedPublished:Apr 21, 2024License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Warm

Sujet Finance 8B v0.1 by sujet-ai is a fine-tuned LLAMA 3 8B language model specialized for financial applications. It excels at financial Yes/No question answering, topic classification across 20 finance-related categories, and sentiment analysis (positive, negative, neutral, bearish, bullish). This model is optimized for accurate and insightful responses to financial queries, leveraging a comprehensive 177k instruction dataset.

Loading preview...

Sujet Finance 8B v0.1: Specialized Financial Language Model

Sujet Finance 8B v0.1, developed by sujet-ai, is a fine-tuned version of the LLAMA 3 8B model, specifically optimized for financial tasks. It was meticulously trained on the proprietary Sujet Finance Instruct-177k dataset, focusing on three core financial applications.

Key Capabilities

  • Yes/No Question Answering: Accurately responds to financial questions requiring a binary 'yes' or 'no' answer.
  • Topic Classification: Categorizes financial texts into 20 specific finance-related classes, including company news, markets, and earnings.
  • Sentiment Analysis: Analyzes financial texts to determine sentiment, classifying it as positive, negative, neutral, bearish, or bullish.

Training Methodology

The model was fine-tuned using LoRA (Low-Rank Adaptation) with specific parameters (r=16, alpha=32) over 1 epoch. A balanced training approach was employed, utilizing a dataset of 17,036 examples to ensure robust performance across diverse financial questions and topics. The model's performance was evaluated against the base LLAMA 3 model, demonstrating impressive results with a strict criterion for correctness (true answer within the first 10 generated words).

Good For

  • Automating financial information extraction.
  • Categorizing financial news and reports.
  • Analyzing market sentiment from textual data.
  • Developing specialized financial chatbots or assistants.

Popular Sampler Settings

Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.

temperature
top_p
top_k
frequency_penalty
presence_penalty
repetition_penalty
min_p