binary-husky/gpt_academic · error · ValueError

AZURE_CFG_ARRAY中配置的模型必须以azure开头

Error message

AZURE_CFG_ARRAY中配置的模型必须以azure开头

What it means

At import time, bridge_all.py iterates the AZURE_CFG_ARRAY config (from shared_utils/config_loader via get_conf) and requires every key (model name) to start with 'azure'; otherwise it raises ValueError. This is a naming contract: azure-prefixed names are what route requests to the Azure endpoint wiring defined right below the check.

Source

Thrown at request_llms/bridge_all.py:1383

        continue
    model_info.update({
        model: {
            "fn_with_ui": ollama_ui,
            "fn_without_ui": ollama_noui,
            "endpoint": ollama_endpoint,
            "max_token": max_token_tmp,
            "tokenizer": tokenizer_gpt35,
            "token_cnt": get_token_num_gpt35,
        },
    })

# -=-=-=-=-=-=- azure模型对齐支持 -=-=-=-=-=-=-
AZURE_CFG_ARRAY = get_conf("AZURE_CFG_ARRAY") # <-- 用于定义和切换多个azure模型 -->
if len(AZURE_CFG_ARRAY) > 0:
    for azure_model_name, azure_cfg_dict in AZURE_CFG_ARRAY.items():
        # 可能会覆盖之前的配置，但这是意料之中的
        if not azure_model_name.startswith('azure'):
            raise ValueError("AZURE_CFG_ARRAY中配置的模型必须以azure开头")
        endpoint_ = azure_cfg_dict["AZURE_ENDPOINT"] + \
            f'openai/deployments/{azure_cfg_dict["AZURE_ENGINE"]}/chat/completions?api-version=2023-05-15'
        model_info.update({
            azure_model_name: {
                "fn_with_ui": chatgpt_ui,
                "fn_without_ui": chatgpt_noui,
                "endpoint": endpoint_,
                "azure_api_key": azure_cfg_dict["AZURE_API_KEY"],
                "max_token": azure_cfg_dict["AZURE_MODEL_MAX_TOKEN"],
                "tokenizer": tokenizer_gpt35,   # tokenizer只用于粗估token数量
                "token_cnt": get_token_num_gpt35,
            }
        })
        if azure_model_name not in AVAIL_LLM_MODELS:
            AVAIL_LLM_MODELS += [azure_model_name]

# -=-=-=-=-=-=- Openrouter模型对齐支持 -=-=-=-=-=-=-
# 为了更灵活地接入Openrouter路由，设计了此接口

View on GitHub (pinned to d6bde0fa54)

Solutions

1. Rename every key in AZURE_CFG_ARRAY to start with lowercase 'azure', e.g. 'azure-gpt-4-32k' (then select that name as LLM_MODEL).
2. Check for case/whitespace: 'Azure-xxx' and ' azure-xxx' both fail the startswith check.
3. Restart the app after fixing config — the check runs at import time, not per request.
4. If you intended a non-Azure model, remove it from AZURE_CFG_ARRAY and configure it under the regular model section instead.

Example fix

# before (config)
AZURE_CFG_ARRAY = {
  'gpt-4-azure': {'AZURE_ENDPOINT': ..., 'AZURE_API_KEY': ..., 'AZURE_ENGINE': ...}
}

# after
AZURE_CFG_ARRAY = {
  'azure-gpt-4': {'AZURE_ENDPOINT': ..., 'AZURE_API_KEY': ..., 'AZURE_ENGINE': ...}
}

def validate_azure_cfg(cfg: dict) -> list[str]:
    return [name for name in cfg if not name.startswith('azure')]

bad = validate_azure_cfg(get_conf('AZURE_CFG_ARRAY'))
if bad:
    raise SystemExit(f'rename {bad} to start with "azure" in AZURE_CFG_ARRAY')

try:
    import request_llms.bridge_all  # noqa: F401  (config check runs at import)
except ValueError as e:
    if 'azure开头' in str(e):
        raise SystemExit(f'config error: {e} — fix AZURE_CFG_ARRAY keys and restart')
    raise

When it happens

Trigger: Editing config so that AZURE_CFG_ARRAY contains a key like 'gpt-4-32k' or 'my_azure_model' (no 'azure' prefix) — the import of request_llms.bridge_all then fails immediately, usually at startup or when any module first imports the bridge.

Common situations: Users copy an Azure model block from another project keeping the original model name; typos like 'Azure-gpt4' (capital A — startswith('azure') is case-sensitive and fails); commenting out one field but renaming the key; config shared across machines with local edits.

Related errors

• Endpoint不正确, 请检查AZURE_ENDPOINT的配置! 当前的Endpoint为:
• AZURE_CFG_ARRAY中配置的模型必须以azure开头
• OpenAI拒绝了请求:
• OpenAI拒绝了请求：
• 由于提问含不合规内容被过滤。

AI-assisted analysis of binary-husky/gpt_academic@d6bde0fa54 (2026-08-14). Data as JSON: /api/errors/ecda302df0e80315. Report an issue: GitHub.