Hugging Face Model Overview
anthracite-org/magnum-v4-72b is a 72.7 billion parameter language model developed by anthracite-org, built upon the Qwen2.5-72B-Instruct base model. Its primary objective is to emulate the advanced prose quality found in Claude 3 models, specifically Sonnet and Opus, making it highly effective for sophisticated text generation.
Key Capabilities
- High-Quality Prose Generation: Fine-tuned to replicate the nuanced and high-quality writing style of Claude 3 models.
- Extensive Context Window: Supports a large context length of 131072 tokens, enabling it to handle long and complex interactions or documents.
- Instruction Following: Benefits from its base on an instruct-tuned model, allowing for robust adherence to user prompts and system instructions.
- ChatML Format: Utilizes the ChatML format for prompting, ensuring structured and clear conversational turns.
Training Details
The model underwent full-parameter fine-tuning using 8x mi300x GPUs. The training leveraged a diverse set of datasets, including anthracite-org/c2_logs_32k_llama3_qwen2_v1.2, anthracite-org/kalo-opus-instruct-22k-no-refusal, and anthracite-org/nopm_claude_writing_fixed, among others, all formatted for ChatML conversations. This extensive dataset curation focused on enhancing its conversational abilities and prose generation.
Use Cases
This model is particularly well-suited for applications requiring advanced conversational capabilities, creative writing, role-playing, and any task where generating human-like, high-quality, and contextually rich text is paramount. Its large context window also makes it ideal for processing and generating long-form content.