CodeQwen1.5Open Weight

Released

CodeQwen1.5 is an AI model released by Qwen on Tuesday, Apr 16 2024, 19 days after Qwen1.5-MoE. It is an open-weight model — the trained weights are available to download and run. It is a 7B parameterThe learned values that shape how a model responds. B means billion and T means trillion. For mixture-of-experts models, the total count is larger than the number active for each token. Parameter count alone does not measure quality. model with a 64k token context windowHow much information the model can work with in one request, measured in tokens — pieces of words or other data. This includes your question, attachments, conversation and reply. Limits can vary by model variant and provider.. Benchmark results (shown below) cover HumanEval.

Benchmarks

HumanEvalFunction synthesis — 164 small Python problems: the AI is given a function's description and has to write the working function. This was the coding benchmark of the GPT-3.5 and GPT-4 era, before the field moved to fixing real bugs in real repositories. Higher is better.
83.5%
#38 of 88

Compare CodeQwen1.5 with

CodeQwen1.5

About

CodeQwen1.5 was released on April 16, 2024 as the first dedicated Qwen coding model. Alibaba published base and chat checkpoints trained on roughly three trillion code-related tokens across 92 programming languages.

With 64K context, the model was designed to work with longer code inputs, including editing and SQL tasks. The chat checkpoint scored 83.5 on zero-shot HumanEval in Qwen's launch tests. It was the predecessor to the multi-size Qwen2.5-Coder family.

Frequently asked questions

CodeQwen1.5 was released by Qwen on Tuesday, Apr 16 2024.

All Qwen releases

42 tracked