关于调用千帆api
请问如何更新model列表?像一些比较新的模型如ERNIE 4.0\3.5 都不在qianfan支持的模型列表内。
qianfan模型代码是这个metagpt/provider/qianfan_api.py。
是模型调不通还是说扣费结算问题?
如果是扣费结算问题,可以在metagpt/utils/token_counter.py中添加:
"""
QianFan Token Price https://cloud.baidu.com/doc/WENXINWORKSHOP/s/hlrk4akp7#tokens%E5%90%8E%E4%BB%98%E8%B4%B9
Due to QianFan has multi price strategies, we unify `Tokens post-payment` as a statistical method.
"""
QIANFAN_MODEL_TOKEN_COSTS = {
"ERNIE-Bot-4": {"prompt": 0.017, "completion": 0.017},
"ERNIE-Bot-8k": {"prompt": 0.0034, "completion": 0.0067},
"ERNIE-Bot": {"prompt": 0.0017, "completion": 0.0017},
"ERNIE-Bot-turbo": {"prompt": 0.0011, "completion": 0.0011},
"EB-turbo-AppBuilder": {"prompt": 0.0011, "completion": 0.0011},
"ERNIE-Speed": {"prompt": 0.00056, "completion": 0.0011},
"BLOOMZ-7B": {"prompt": 0.00056, "completion": 0.00056},
"Llama-2-7B-Chat": {"prompt": 0.00056, "completion": 0.00056},
"Llama-2-13B-Chat": {"prompt": 0.00084, "completion": 0.00084},
"Llama-2-70B-Chat": {"prompt": 0.0049, "completion": 0.0049},
"ChatGLM2-6B-32K": {"prompt": 0.00056, "completion": 0.00056},
"AquilaChat-7B": {"prompt": 0.00056, "completion": 0.00056},
"Mixtral-8x7B-Instruct": {"prompt": 0.0049, "completion": 0.0049},
"SQLCoder-7B": {"prompt": 0.00056, "completion": 0.00056},
"CodeLlama-7B-Instruct": {"prompt": 0.00056, "completion": 0.00056},
"XuanYuan-70B-Chat-4bit": {"prompt": 0.0049, "completion": 0.0049},
"Qianfan-BLOOMZ-7B-compressed": {"prompt": 0.00056, "completion": 0.00056},
"Qianfan-Chinese-Llama-2-7B": {"prompt": 0.00056, "completion": 0.00056},
"Qianfan-Chinese-Llama-2-13B": {"prompt": 0.00084, "completion": 0.00084},
"ChatLaw": {"prompt": 0.0011, "completion": 0.0011},
"Yi-34B-Chat": {"prompt": 0.0, "completion": 0.0},
}
QIANFAN_ENDPOINT_TOKEN_COSTS = {
"completions_pro": QIANFAN_MODEL_TOKEN_COSTS["ERNIE-Bot-4"],
"ernie_bot_8k": QIANFAN_MODEL_TOKEN_COSTS["ERNIE-Bot-8k"],
"completions": QIANFAN_MODEL_TOKEN_COSTS["ERNIE-Bot"],
"eb-instant": QIANFAN_MODEL_TOKEN_COSTS["ERNIE-Bot-turbo"],
"ai_apaas": QIANFAN_MODEL_TOKEN_COSTS["EB-turbo-AppBuilder"],
"ernie_speed": QIANFAN_MODEL_TOKEN_COSTS["ERNIE-Speed"],
"bloomz_7b1": QIANFAN_MODEL_TOKEN_COSTS["BLOOMZ-7B"],
"llama_2_7b": QIANFAN_MODEL_TOKEN_COSTS["Llama-2-7B-Chat"],
"llama_2_13b": QIANFAN_MODEL_TOKEN_COSTS["Llama-2-13B-Chat"],
"llama_2_70b": QIANFAN_MODEL_TOKEN_COSTS["Llama-2-70B-Chat"],
"chatglm2_6b_32k": QIANFAN_MODEL_TOKEN_COSTS["ChatGLM2-6B-32K"],
"aquilachat_7b": QIANFAN_MODEL_TOKEN_COSTS["AquilaChat-7B"],
"mixtral_8x7b_instruct": QIANFAN_MODEL_TOKEN_COSTS["Mixtral-8x7B-Instruct"],
"sqlcoder_7b": QIANFAN_MODEL_TOKEN_COSTS["SQLCoder-7B"],
"codellama_7b_instruct": QIANFAN_MODEL_TOKEN_COSTS["CodeLlama-7B-Instruct"],
"xuanyuan_70b_chat": QIANFAN_MODEL_TOKEN_COSTS["XuanYuan-70B-Chat-4bit"],
"qianfan_bloomz_7b_compressed": QIANFAN_MODEL_TOKEN_COSTS["Qianfan-BLOOMZ-7B-compressed"],
"qianfan_chinese_llama_2_7b": QIANFAN_MODEL_TOKEN_COSTS["Qianfan-Chinese-Llama-2-7B"],
"qianfan_chinese_llama_2_13b": QIANFAN_MODEL_TOKEN_COSTS["Qianfan-Chinese-Llama-2-13B"],
"chatlaw": QIANFAN_MODEL_TOKEN_COSTS["ChatLaw"],
"yi_34b_chat": QIANFAN_MODEL_TOKEN_COSTS["Yi-34B-Chat"],
}
感谢您百忙之中的回复!
2025-01-17 17:16:03.811 | ERROR | metagpt.utils.common:log_it:554 - Finished call to 'metagpt.actions.action_node.ActionNode._aask_v1' after 0.000(s), this was the 1st time calling it. exp: The provided model ERNIE-4.0-8Kis not in the list of supported models. If this is a recently added model, try using theendpointarguments and create an issue to tell us. Supported models: {'ERNIE-Bot', 'XuanYuan-70B-Chat-4bit', 'ERNIE-3.5-4K-0205', 'ChatGLM2-6B-32K', 'ERNIE-Bot-8k', 'ERNIE-Bot-4', 'ERNIE-Bot-turbo-AI', 'Qianfan-Chinese-Llama-2-13B', 'ERNIE-Bot-turbo', 'ChatLaw', 'Qianfan-Chinese-Llama-2-7B', 'BLOOMZ-7B', 'Mixtral-8x7B-Instruct', 'Llama-2-70b-chat', 'ERNIE-Speed', 'Llama-2-7b-chat', 'EB-turbo-AppBuilder', 'Llama-2-13b-chat', 'ERNIE-3.5-8K-0205', 'AquilaChat-7B', 'Qianfan-BLOOMZ-7B-compressed', 'Yi-34B-Chat'}
问题如上,我在调用最新的ERNIE-4.0-8K模型时,显示``ERNIE-4.0-8K is not in the list of supported models,请问是直接在metagpt/provider/qianfan_api.py中添加模型名称就可以正常调用吗?
This issue has no activity in the past 30 days. Please comment on the issue if you have anything to add.
This issue was closed due to 45 days of inactivity. If you feel this issue is still relevant, please reopen the issue to continue the discussion.