mdba/escutia-qlora

TEXT GENERATIONPricing:Input $0.04 / Cached $0.008 / Output $0.08Concurrent Unit Cost:1Model Size:1.5BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Aug 26, 2026Architecture:Transformer Featherless Exclusive Cold

mdba/escutia-qlora is a 1.5 billion parameter causal language model, fine-tuned from Qwen/Qwen2.5-1.5B-Instruct. Developed by mdba, this model is specifically optimized for sentiment classification tasks in Portuguese. It leverages QLoRA training for efficient adaptation and offers a 32768 token context length, making it suitable for analyzing longer texts.

Loading preview...

EscutIA QLoRA: Portuguese Sentiment Classification

This model, developed by mdba, is a specialized version of the Qwen/Qwen2.5-1.5B-Instruct base model, fine-tuned using the QLoRA method. With 1.5 billion parameters and a substantial 32768 token context length, it is engineered for efficient and accurate sentiment analysis in the Portuguese language.

Key Capabilities

  • Portuguese Sentiment Classification: Specifically trained and optimized to identify and categorize sentiment within Portuguese text.
  • QLoRA Fine-tuning: Benefits from efficient QLoRA training, allowing for effective adaptation to its specialized task while maintaining a compact size.
  • Full Weights Provided: The repository contains the complete model weights, enabling direct loading and use with the Hugging Face transformers library without needing separate PEFT adapters.

Good For

  • Analyzing sentiment in Portuguese customer reviews, social media, or news articles.
  • Applications requiring a focused, performant model for Portuguese text understanding.
  • Developers looking for a ready-to-use solution for sentiment analysis in Portuguese.

Usage Example

Loading the model is straightforward using the transformers library:

from transformers import AutoModelForCausalLM, AutoTokenizer
model_id = "mdba/escutia-qlora"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto")