Stage-jh-monitor/qwen35-4b-filter-s_signal5-200-qwen38-27b-newprompt-4k-epoch3

VISIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:4.5BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Sep 15, 2026Architecture:Transformer Featherless Exclusive Cold

Stage-jh-monitor/qwen35-4b-filter-s_signal5-200-qwen38-27b-newprompt-4k-epoch3 is a 4.5 billion parameter language model based on the Qwen 3.5 architecture, fine-tuned using a reinforcement learning (RL) method. It was trained on the Stage-org/qwen35-4b-filter-s_signal5-200-qwen38-27b-newprompt-4k dataset with a context length of 32768 tokens. This model is optimized for tasks requiring advanced reasoning and tool-use capabilities, leveraging specific parsing for Qwen3 reasoning and tool calls.

Loading preview...

Overview

Stage-jh-monitor/qwen35-4b-filter-s_signal5-200-qwen38-27b-newprompt-4k-epoch3 is a 4.5 billion parameter language model derived from the Qwen 3.5 architecture. It has undergone reinforcement learning (RL) fine-tuning over 3 epochs, utilizing a substantial dataset for enhanced performance. The model supports an extended context length of 32768 tokens, making it suitable for processing longer inputs and generating comprehensive responses.

Key Capabilities

  • Reinforcement Learning Fine-tuning: Trained with an RL method, specifically using a DPO-like loss function with masking, to optimize for specific task performance.
  • Extended Context Window: Features a max_model_len of 65536 tokens for inference, allowing for deep contextual understanding and generation.
  • Advanced Reasoning and Tool Use: Configured with reasoning_parser="qwen3" and tool_call_parser="qwen3_coder", indicating specialized capabilities for complex reasoning tasks and integrating with external tools.
  • Optimized Inference: Utilizes flash_attention_2 for efficient attention mechanisms and vllm_extra for language model-only inference, ensuring high throughput.

Good For

  • Complex Reasoning Tasks: Ideal for applications requiring sophisticated logical deduction and problem-solving, leveraging its Qwen3-specific reasoning parser.
  • Tool-Augmented Generation: Suitable for scenarios where the model needs to interact with external tools or APIs, facilitated by its Qwen3 coder tool-call parser.
  • Long-Context Applications: Excellent for tasks that benefit from processing extensive textual information, such as document analysis, summarization of large texts, or maintaining long-running conversations.