NewEden/Gemma-27B-chatml

Hugging Face
VISIONPricing:Input $0.4 / Cached $0.08 / Output $1.2Concurrent Unit Cost:2Model Size:27BQuant:FP8Context Size:32kPublished:Apr 16, 2025License:gemmaArchitecture:Transformer Featherless Exclusive Warm

NewEden/Gemma-27B-chatml is a 27 billion parameter multimodal model from Google DeepMind, part of the Gemma 3 family, capable of processing both text and image inputs to generate text outputs. This instruction-tuned variant features a large 128K token context window and multilingual support for over 140 languages. It excels in diverse tasks such as question answering, summarization, reasoning, and image understanding, making it suitable for deployment in resource-constrained environments.

Loading preview...

Gemma 3: A Multimodal, Multilingual Model Family

Gemma 3, developed by Google DeepMind, is a family of open, lightweight models built with the same research and technology as the Gemini models. The NewEden/Gemma-27B-chatml variant is a 27 billion parameter instruction-tuned model designed for multimodal interactions, accepting both text and image inputs to produce text outputs.

Key Capabilities and Features

  • Multimodal Input: Processes text strings and images (normalized to 896x896 resolution, encoded to 256 tokens each).
  • Large Context Window: Features a substantial 128K token input context for the 27B model, enabling processing of extensive inputs.
  • Multilingual Support: Offers broad language coverage, supporting over 140 languages in its training data.
  • Diverse Task Performance: Well-suited for a range of applications including question answering, summarization, reasoning, and image content analysis.
  • Optimized for Deployment: Its relatively efficient design allows for deployment in environments with limited resources, such as laptops or local cloud infrastructure.
  • Robust Training: Trained on 14 trillion tokens, including web documents, code, mathematics, and images, ensuring broad knowledge and capability.

Performance Highlights (Gemma 3 PT 27B)

  • Reasoning: Achieves 85.6 on HellaSwag (10-shot), 82.4 on BoolQ (0-shot), and 77.7 on BIG-Bench Hard (few-shot).
  • STEM & Code: Scores 78.6 on MMLU (5-shot), 82.6 on GSM8K (8-shot), and 48.8 on HumanEval (0-shot).
  • Multilingual: Demonstrates strong performance with 74.3 on MGSM and 75.7 on Global-MMLU-Lite.
  • Multimodal: Achieves 116 on COCOcap, 85.6 on DocVQA, and 72.9 on VQAv2.

Intended Use Cases

  • Content Creation: Generating creative text formats, marketing copy, and email drafts.
  • Conversational AI: Powering chatbots and virtual assistants.
  • Text Summarization: Creating concise summaries of documents and reports.
  • Image Data Extraction: Interpreting and summarizing visual data for text communications.
  • Research and Education: Serving as a foundation for VLM/NLP research and language learning tools.

Finetunes
1 models