localized-ft/Qwen3-32B-target-only-no-hallucination-last-third-sft-bf16
TEXT GENERATIONPricing:Input $0.408 / Cached $0.0816 / Output $1.972Concurrent Unit Cost:2Model Size:32BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 27, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold
The localized-ft/Qwen3-32B-target-only-no-hallucination-last-third-sft-bf16 is a 32 billion parameter Qwen3 model developed by localized-ft. It was finetuned using Unsloth and Huggingface's TRL library, enabling 2x faster training. This model is specifically optimized to reduce hallucinations and target specific outputs, making it suitable for applications requiring high factual accuracy and controlled generation within its 32768 token context length.
Loading preview...
Model Overview
This model, developed by localized-ft, is a 32 billion parameter Qwen3 variant that has been finetuned from unsloth/Qwen3-32B. The finetuning process leveraged Unsloth and Huggingface's TRL library, which facilitated a 2x faster training speed compared to standard methods.
Key Characteristics
- Architecture: Qwen3-32B base model.
- Parameter Count: 32 billion parameters.
- Context Length: Supports a context window of 32768 tokens.
- Training Optimization: Utilizes Unsloth for accelerated finetuning.
- Focus: Specifically trained to minimize hallucinations and target precise outputs, enhancing reliability for specific use cases.
Ideal Use Cases
- Applications requiring high factual accuracy and reduced generative errors.
- Scenarios where controlled and targeted text generation is critical.
- Tasks benefiting from a large 32B parameter model with an extended context window.
- Developers seeking a finetuned Qwen3 model with optimized training efficiency.