mirror of
https://github.com/discourse/discourse-ai.git
synced 2025-03-09 02:40:50 +00:00
0634b85a81
* UX: Validations to Llm-backed features (except AI Bot) This change is part of an ongoing effort to prevent enabling a broken feature due to lack of configuration. We also want to explicit which provider we are going to use. For example, Claude models are available through AWS Bedrock and Anthropic, but the configuration differs. Validations are: * You must choose a model before enabling the feature. * You must turn off the feature before setting the model to blank. * You must configure each model settings before being able to select it. * Add provider name to summarization options * vLLM can technically support same models as HF * Check we can talk to the selected model * Check for Bedrock instead of anthropic as a site could have both creds setup
66 lines
1.6 KiB
Ruby
66 lines
1.6 KiB
Ruby
# frozen_string_literal: true
|
|
|
|
module DiscourseAi
|
|
module Completions
|
|
module Endpoints
|
|
class CannedResponse
|
|
CANNED_RESPONSE_ERROR = Class.new(StandardError)
|
|
|
|
def self.can_contact?(_, _)
|
|
Rails.env.test?
|
|
end
|
|
|
|
def initialize(responses)
|
|
@responses = responses
|
|
@completions = 0
|
|
@prompt = nil
|
|
end
|
|
|
|
def normalize_model_params(model_params)
|
|
# max_tokens, temperature, stop_sequences are already supported
|
|
model_params
|
|
end
|
|
|
|
attr_reader :responses, :completions, :prompt
|
|
|
|
def perform_completion!(prompt, _user, _model_params)
|
|
@prompt = prompt
|
|
response = responses[completions]
|
|
if response.nil?
|
|
raise CANNED_RESPONSE_ERROR,
|
|
"The number of completions you requested exceed the number of canned responses"
|
|
end
|
|
|
|
@completions += 1
|
|
if block_given?
|
|
cancelled = false
|
|
cancel_fn = lambda { cancelled = true }
|
|
|
|
# We buffer and return tool invocations in one go.
|
|
if is_tool?(response)
|
|
yield(response, cancel_fn)
|
|
else
|
|
response.each_char do |char|
|
|
break if cancelled
|
|
yield(char, cancel_fn)
|
|
end
|
|
end
|
|
else
|
|
response
|
|
end
|
|
end
|
|
|
|
def tokenizer
|
|
DiscourseAi::Tokenizer::OpenAiTokenizer
|
|
end
|
|
|
|
private
|
|
|
|
def is_tool?(response)
|
|
Nokogiri::HTML5.fragment(response).at("function_calls").present?
|
|
end
|
|
end
|
|
end
|
|
end
|
|
end
|