mirror of
https://github.com/BerriAI/litellm.git
synced 2026-10-12 03:48:56 +00:00
2f933c3f59
Closes the documented prefill-API coverage gap (#4881) for the hosted_vllm provider so /v1/messages requests routed to vLLM honor the Anthropic prefill semantic ("when the last message is role:assistant, the model continues from where that message leaves off"). Two minimal changes: * `litellm/llms/hosted_vllm/chat/transformation.py`: extend transform_request and async_transform_request to detect the unified prefill marker (`prefix: True` on the trailing assistant message) and inject vLLM's native continuation flags into extra_body — `continue_final_message: true` and `add_generation_prompt: false`. Idempotent (setdefault), no-op when the trigger is absent, never overrides caller-set values. * `litellm/llms/anthropic/experimental_pass_through/adapters/handler.py`: in `_prepare_completion_kwargs` — which is reached only when /v1/messages is routed to a non-Anthropic provider — auto-tag a trailing role:"assistant" message with `prefix: True` so the provider-side translator above fires. This makes the Anthropic /v1/messages prefill contract portable across providers. Tests added in test_hosted_vllm_chat_transformation.py: - prefill marker → continue_final_message injected - no marker → no-op - user-ending → no-op - existing extra_body preserved and merged - explicit caller-set continue_final_message:False respected Verified locally: 13/13 hosted_vllm transformation tests pass (8 pre-existing + 5 new). Refs: - #4881 — closed; defines the unified prefill API but only Anthropic provider implemented it - BerriAI/litellm#28580 — the bug report with empirical trace |
||
|---|---|---|
| .. | ||
| agent_tests | ||
| audio_tests | ||
| basic_proxy_startup_tests | ||
| batches_tests | ||
| benchmarks | ||
| code_coverage_tests | ||
| documentation_tests | ||
| enterprise | ||
| guardrails_tests | ||
| image_gen_tests | ||
| litellm | ||
| litellm-proxy-extras | ||
| litellm_core_utils | ||
| litellm_utils_tests | ||
| llm_responses_api_testing | ||
| llm_translation | ||
| load_tests | ||
| local_testing | ||
| logging_callback_tests | ||
| mcp_tests | ||
| multi_instance_e2e_tests | ||
| ocr_tests | ||
| old_proxy_tests/tests | ||
| openai_endpoints_tests | ||
| otel_tests | ||
| pass_through_tests | ||
| pass_through_unit_tests | ||
| proxy_admin_ui_tests | ||
| proxy_e2e_anthropic_messages_tests | ||
| proxy_security_tests | ||
| proxy_unit_tests | ||
| router_unit_tests | ||
| scim_tests | ||
| search_tests | ||
| spend_tracking_tests | ||
| store_model_in_db_tests | ||
| test_litellm | ||
| unified_google_tests | ||
| vector_store_tests | ||
| windows_tests | ||
| __init__.py | ||
| _flush_vcr_cache.py | ||
| _vcr_conftest_common.py | ||
| _vcr_redis_persister.py | ||
| eval_swe_bench.py | ||
| gettysburg.wav | ||
| large_text.py | ||
| openai_batch_completions.jsonl | ||
| README.MD | ||
| test_budget_management.py | ||
| test_callbacks_on_proxy.py | ||
| test_config.py | ||
| test_debug_warning.py | ||
| test_default_encoding_non_root.py | ||
| test_end_users.py | ||
| test_entrypoint.py | ||
| test_fallbacks.py | ||
| test_gpt5_azure_temperature_support.py | ||
| test_health.py | ||
| test_keys.py | ||
| test_litellm_proxy_responses_config.py | ||
| test_logging.conf | ||
| test_models.py | ||
| test_new_vector_store_endpoints.py | ||
| test_openai_endpoints.py | ||
| test_organizations.py | ||
| test_otel_thread_leak.py | ||
| test_passthrough_endpoints.py | ||
| test_presidio_latency.py | ||
| test_proxy_server_non_root.py | ||
| test_ratelimit.py | ||
| test_resource_cleanup.py | ||
| test_service_logger_otel.py | ||
| test_spend_logs.py | ||
| test_team.py | ||
| test_team_logging.py | ||
| test_team_members.py | ||
| test_users.py | ||
In total litellm runs 1000+ tests
[02/20/2025] Update:
To make it easier to contribute and map what behavior is tested,
we've started mapping the litellm directory in tests/test_litellm
This folder can only run mock tests.