Module: Legion::Extensions::Llm::Vllm::ProviderManagementMethods
- Included in:
- Provider
- Defined in:
- lib/legion/extensions/llm/vllm/provider.rb
Overview
Public methods for vLLM management/admin endpoints. Included into Provider to keep the class body within length limits.
Instance Method Summary collapse
- #reset_mm_cache ⇒ Object
- #reset_prefix_cache(reset_running_requests: nil, reset_external: nil) ⇒ Object
- #sleep(level: 1) ⇒ Object
- #version ⇒ Object
- #wake_up(tags: nil) ⇒ Object
Instance Method Details
#reset_mm_cache ⇒ Object
29 30 31 32 |
# File 'lib/legion/extensions/llm/vllm/provider.rb', line 29 def reset_mm_cache log.debug { 'resetting vLLM multimodal cache' } connection.post(reset_mm_cache_url, {}).body end |
#reset_prefix_cache(reset_running_requests: nil, reset_external: nil) ⇒ Object
21 22 23 24 25 26 27 |
# File 'lib/legion/extensions/llm/vllm/provider.rb', line 21 def reset_prefix_cache(reset_running_requests: nil, reset_external: nil) log.debug do "resetting vLLM prefix cache reset_running_requests=#{reset_running_requests.inspect} " \ "reset_external=#{reset_external.inspect}" end connection.post(with_query(reset_prefix_cache_url, reset_running_requests:, reset_external:), {}).body end |
#sleep(level: 1) ⇒ Object
34 35 36 37 |
# File 'lib/legion/extensions/llm/vllm/provider.rb', line 34 def sleep(level: 1) log.debug { "putting vLLM worker to sleep level=#{level.inspect}" } connection.post(with_query(sleep_url, level:), {}).body end |
#version ⇒ Object
16 17 18 19 |
# File 'lib/legion/extensions/llm/vllm/provider.rb', line 16 def version log.info { "fetching version from #{api_base}#{version_url}" } connection.get(version_url).body end |
#wake_up(tags: nil) ⇒ Object
39 40 41 42 43 |
# File 'lib/legion/extensions/llm/vllm/provider.rb', line 39 def wake_up(tags: nil) log.debug { "waking vLLM worker tags=#{Array().inspect}" } query = Array().map { |tag| ['tags', tag] } connection.post(with_query(wake_up_url, query), {}).body end |