license, base_model, language, pipeline_tag, library_name, tags
| license | base_model | language | pipeline_tag | library_name | tags | |||
|---|---|---|---|---|---|---|---|---|
| apache-2.0 |
|
|
text-generation | transformers |
|
MemOperator-4B-f32-GGUF
MemOperator-4B by MemTensor is a specialized causal language model designed for efficient memory operations within the MemOS system. It excels in memory extraction, integration, and updating while enabling local-only deployment for environments without internet access. Derived from the Qwen3-4B architecture and fine-tuned via supervised learning on both human-annotated and generated data, this 4 billion parameter model supports both English and Chinese, and processes long contexts up to 32,768 tokens.
It offers fast, low-resource memory management that outperforms comparably sized open models like GPT-4o-mini, making it ideal for real-time, cost-effective memory tasks in conversational and document settings. MemOperator-4B is designed to seamlessly extract high-quality memories and organize them for enhanced long-term coherence in applications such as MemOS, supporting memory-centric AI workflows with strong multilingual capabilities and robust system performance.
Model Files
| Model File name | Size | QuantType |
|---|---|---|
| MemOperator-4B.BF16.gguf | 8.05 GB | BF16 |
| MemOperator-4B.F16.gguf | 8.05 GB | F16 |
| MemOperator-4B.F32.gguf | 16.1 GB | F32 |
| MemOperator-4B.Q2_K.gguf | 1.67 GB | Q2_K |
| MemOperator-4B.Q3_K_L.gguf | 2.24 GB | Q3_K_L |
| MemOperator-4B.Q3_K_M.gguf | 2.08 GB | Q3_K_M |
| MemOperator-4B.Q3_K_S.gguf | 1.89 GB | Q3_K_S |
| MemOperator-4B.Q4_K_M.gguf | 2.5 GB | Q4_K_M |
| MemOperator-4B.Q4_K_S.gguf | 2.38 GB | Q4_K_S |
| MemOperator-4B.Q5_K_M.gguf | 2.89 GB | Q5_K_M |
| MemOperator-4B.Q5_K_S.gguf | 2.82 GB | Q5_K_S |
| MemOperator-4B.Q6_K.gguf | 3.31 GB | Q6_K |
| MemOperator-4B.Q8_0.gguf | 4.28 GB | Q8_0 |
Quants Usage
(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
Here is a handy graph by ikawrakow comparing some lower-quality quant types (lower is better):
