saas/qianfan 支持 ernie-3.5-128k

[kgbook]

qianfan.errors.InvalidArgumentError: The provided model ERNIE-3.5-128K is not in the list of supported models. If this is a recently added model, try using the endpoint arguments and create an issue to tell us. Supported models: {'ERNIE-Bot-turbo-AI', 'ERNIE-3.5-4K-0205', 'XuanYuan-70B-Chat-4bit', 'ChatGLM2-6B-32K', 'ERNIE-Functions-8K', 'ERNIE Speed-AppBuilder', 'Qianfan-BLOOMZ-7B-compressed', 'Qianfan-Chinese-Llama-2-13B', 'Yi-34B-Chat', 'ERNIE-3.5-8K-preview', 'ERNIE-Tiny-8K', 'ERNIE-Character-8K', 'Llama-2-70B-Chat', 'Llama-2-7B-Chat', 'ERNIE-Speed-128K', 'Mixtral-8x7B-Instruct', 'ERNIE Speed', 'ERNIE-4.0-preemptible', 'Meta-Llama-3-8B', 'Qianfan-Chinese-Llama-2-7B', 'Gemma-7B-it', 'ERNIE-3.5-8K-preemptible', 'ERNIE-Bot', 'BLOOMZ-7B', 'EB-turbo-AppBuilder', 'ERNIE-3.5-8K', 'ERNIE 3.5', 'ERNIE-Lite-8K', 'AquilaChat-7B', 'ERNIE-Speed', 'ERNIE-4.0-8K', 'Llama-2-13B-Chat', 'ERNIE-Speed-8K', 'ERNIE-Bot-8K', 'ERNIE-Lite-8K-0308', 'ERNIE-Bot-4', 'ERNIE-4.0-preview', 'ERNIE-3.5-8K-0205', 'ERNIE-Bot-turbo', 'ERNIE-Lite-8K-0922', 'ChatLaw', 'Meta-Llama-3-70B', 'ERNIE-3.5-8K-1222'}

2024年5月16日发布 ERNIE-3.5-128K，提升上下文窗口长度至128K

endpoint: /chat/ernie-3.5-128k

详细信息如下：

"ERNIE-3.5-128K": QfLLMInfo(
                endpoint="/chat/ernie-3.5-128k",
                required_keys={"messages"},
                optional_keys={
                    "stream",
                    "temperature",
                    "top_p",
                    "penalty_score",
                    "functions",
                    "system",
                    "user_id",
                    "user_setting",
                    "stop",
                    "disable_search",
                    "enable_citation",
                    "max_output_tokens",
                    "response_format",
                    "tool_choice",
                    "enable_trace",
                },
                max_input_chars=516096,
                max_input_tokens=126976,
                input_price_per_1k_tokens=0.012,
                output_price_per_1k_tokens=0.012,

官方文档： https://cloud.baidu.com/doc/WENXINWORKSHOP/s/dlw4ptsq7