Closed actow closed 2 weeks ago
More References:
https://github.com/ollama/ollama/blob/main/templates/chatml.gotmpl
Hi. You can manually create such a template.
BOS = false, EOS = false
[system](<|im_start|>system
You are a helpful assistant.)
<eos><|im_start|>user
{prompt}<eos><|im_start|>assistant
stop words = <|im_start|>,<eos>
My template:
BOS = false, EOS = false, Special = true
Format:
<|im_start|>system
You are a helpful assistant.
<eos>
<|im_start|>user
{prompt}<eos>
<|im_start|>assistant
Skip tokens:
<|im_start|>,<eos>
Metal = true
iPhone12pro Qwen2-1.5B-Instruct-Q8_0 works fine with conversation and codegen.
My config json: Qwen2.json
thanks a lot. It will be even better if that becomes a built-in template. Given that Qwen2 is gaining popularity and it has 0.5b and 1.5b models, I'd imagine this can be useful to many people.
Also, is there a way to import templates?
Also, is there a way to import templates?
For iPhone, just put json file in "File" App -> My iPhone -> LLM Farm -> model_setting_templates.
That worked. Thanks a lot.
The Qwen2 models uses ChatML.
See https://huggingface.co/Qwen/Qwen2-7B-Instruct/blob/main/tokenizer_config.json