lex-dataset 0.2.4 → 0.2.6

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: fd78edb40ac84145152b707d558dafa107c0bd9ae45995c4b035e17e55392989
-  data.tar.gz: 61951016c2bc00670d0f58e7157150f50693e7e7da06b863a1bf7085a6a9fc20
+  metadata.gz: 0114e9e266c0a6f2d4d88fefcf535a91bae559db9784d03a775dbdc1c9b95bf9
+  data.tar.gz: 28b59c1e639d99123f54d1f5630fd02df62c9b6912da06679d917dede310e71b
 SHA512:
-  metadata.gz: ca56908b97943c80ddd750383930de3b7700cfcd86ca9e6eae9b5a3af42f44a74df69e5d9f97dfb4544213dc2b41243c01ac3e4956ae56e2cd53169fd480d1bf
-  data.tar.gz: f4dda491c496cffd270a451ca704b8fa2b4ba231d7a6c411ed5c00fc8cb12965a5429a19b0df4b6a7f16f821d291328f9754e501f938da8fb324dcd2710a4b8a
+  metadata.gz: b9be421ed0c919084fd388f151e465f98e08b095e0bf1756358d2ec9bd813d88654a0dc0f60a5fbeb09cd3c1c9c058f10c056ae79011c5bae9a4cea42b862454
+  data.tar.gz: f0ed08bcc4d021ae90d489b3bce2fd1a47629bfe7ea8e55538945afc8ad6a8052f88316452d6d28fcafa69e952a018089b3ede1f42649e0e9216f4e4698afb4d

data/lib/legion/extensions/dataset/runners/dataset.rb

@@ -8,6 +8,12 @@ module Legion
     module Dataset
       module Runners
         module Dataset
+          extend self
+
+          def self.remote_invocable?
+            false
+          end
+
           def create_dataset(name:, description: nil, rows: [], **)
             ds_id = db[:datasets].insert(name: name, description: description, created_at: Time.now.utc)
             create_version(ds_id, rows)
@@ -62,7 +68,7 @@ module Legion
             return { error: 'legion-llm is not available' } unless llm_available?
 
             rows = call_llm_for_rows(description: description, count: count, schema: schema, model: model)
-            return rows if rows.is_a?(Hash) && rows[:error]
+            return rows if rows.is_a?(Hash) && rows[:error] # rubocop:disable Legion/Extension/RunnerReturnHash
 
             result = create_dataset(name: name, description: description, rows: rows)
             result.merge(generated: true)
@@ -92,14 +98,14 @@ module Legion
 
           def invoke_llm(prompt:, **llm_opts)
             result = if Legion::LLM.respond_to?(:structured)
-                       Legion::LLM.structured(
+                       Legion::LLM.structured( # rubocop:disable Legion/HelperMigration/DirectLlm
                          message: prompt,
                          schema:  generate_schema,
                          caller:  { extension: 'lex-dataset', operation: 'generate' },
                          **llm_opts
                        )
                      else
-                       Legion::LLM.chat(message: prompt, caller: { extension: 'lex-dataset', operation: 'generate' }, **llm_opts)
+                       Legion::LLM.chat(message: prompt, caller: { extension: 'lex-dataset', operation: 'generate' }, **llm_opts) # rubocop:disable Legion/HelperMigration/DirectLlm
                      end
             content = result.respond_to?(:content) ? result.content : result.to_s
             content.strip.sub(/\A```(?:json)?\n?/, '').sub(/\n?```\z/, '')
@@ -107,13 +113,13 @@ module Legion
 
           def parse_llm_rows(content)
             parsed = ::JSON.parse(content)
-            return nil unless parsed.is_a?(Array)
+            return nil unless parsed.is_a?(Array) # rubocop:disable Legion/Extension/RunnerReturnHash
 
             parsed.map do |item|
               h = item.transform_keys(&:to_sym)
               { input: h[:input].to_s, expected_output: h[:expected_output]&.to_s }
             end
-          rescue ::JSON::ParserError
+          rescue ::JSON::ParserError => _e
             nil
           end
 
@@ -126,7 +132,8 @@ module Legion
             if schema
               lines << ''
               lines << 'Schema guidance for inputs and outputs:'
-              lines << "```json\n#{::JSON.generate(schema)}\n```"
+              lines << ''
+              lines << ::JSON.generate(schema)
             end
             lines << ''
             lines << 'Respond ONLY with a valid JSON array, no other text.'

data/lib/legion/extensions/dataset/runners/experiment.rb

@@ -7,6 +7,8 @@ module Legion
     module Dataset
       module Runners
         module Experiment
+          extend self
+
           def run_experiment(name:, dataset_name:, task_callable:, dataset_version: nil, evaluators: [], **)
             ds = get_dataset(name: dataset_name, version: dataset_version)
             return { error: ds[:error] } if ds[:error]
@@ -69,7 +71,7 @@ module Legion
 
           def load_experiment_results(name)
             exp = db[:experiments].where(name: name).first
-            return nil unless exp
+            return nil unless exp # rubocop:disable Legion/Extension/RunnerReturnHash
 
             db[:experiment_results].where(experiment_id: exp[:id]).order(:row_index).all
           end

data/lib/legion/extensions/dataset/runners/sampling.rb

@@ -5,6 +5,8 @@ module Legion
     module Dataset
       module Runners
         module Sampling
+          extend self
+
           def sample_from_traces(dataset_name:, source: :legion_data, filters: {},
                                  sample_size: nil, strategy: :recent, **)
             traces = fetch_traces(source, filters)
@@ -52,7 +54,7 @@ module Legion
 
           def sample_error_biased(traces, size)
             errors, successes = traces.partition { |t| t[:status] == 'error' }
-            return traces unless size
+            return traces unless size # rubocop:disable Legion/Extension/RunnerReturnHash
 
             half = size / 2
             (errors.first(half) + successes.first(size - half)).first(size)
@@ -60,7 +62,7 @@ module Legion
 
           def sample_stratified(traces, size)
             groups = traces.group_by { |t| t[:span_kind] }
-            return traces unless size
+            return traces unless size # rubocop:disable Legion/Extension/RunnerReturnHash
 
             per_group = [size / [groups.size, 1].max, 1].max
             groups.values.flat_map { |g| g.first(per_group) }.first(size)

data/lib/legion/extensions/dataset/version.rb

@@ -3,7 +3,7 @@
 module Legion
   module Extensions
     module Dataset
-      VERSION = '0.2.4'
+      VERSION = '0.2.6'
     end
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: lex-dataset
 version: !ruby/object:Gem::Version
-  version: 0.2.4
+  version: 0.2.6
 platform: ruby
 authors:
 - Matthew Iverson