query
Function
Generates an image from given text prompts.
Prototype
def query(text, llm_config, *args, **kwargs)
Parameters
Parameter |
Data Type |
Required/Optional |
Description |
|---|---|---|---|
text |
String |
Required |
Prompt for generating an image. The value cannot be empty and can contain a maximum of 1024 × 1024 characters. |
llm_config |
LLMParameterConfig |
Optional |
Parameters for calling an LLM. For details, see LLMParameterConfig. |
args |
List |
Optional |
(Not used) Inherited from the base class. |
kwargs["output_format"] |
String |
Optional |
Output image format, which can be obtained by using kwargs["output_format"]. The value can be png, jpeg, jpg, or webp. The default value is png. |
kwargs["size"] |
String |
Optional |
Image size, in the format of "height × width", which is specified by kwargs. The supported size depends on the corresponding LLM. The regular expression matching format is ^\d{1,5}\*\d{1,5}$. The default value is 512 × 512. |
Return Value
Data Type |
Description |
|---|---|
Dict, {"prompt": prompt, "result": data} |
data is the Base64-encoded image data. |
Example
from mx_rag.chain import Text2ImgChain
from mx_rag.llm import Text2ImgMultiModel
from mx_rag.utils import ClientParam
client_param = ClientParam(ca_file="/path/to/ca.crt")
multi_model=Text2ImgMultiModel(model_name="sd", url="text to img url", client_param=client_param)
text2img_chain = Text2ImgChain(multi_model=multi_model)
llm_data = text2img_chain.query("dog wearing black glasses", output_format="jpg")
print(llm_data)