diff options
Diffstat (limited to 'config.yaml')
| -rw-r--r-- | config.yaml | 65 |
1 files changed, 40 insertions, 25 deletions
diff --git a/config.yaml b/config.yaml index ec0326c..637364c 100644 --- a/config.yaml +++ b/config.yaml @@ -5,11 +5,14 @@ defaults: temperature: 1.0 # max_tokens: 4096 - oversight: low + oversight: high include_reasoning_in_context: true # Provider configurations providers: + groq: + api_key_env: GROQ_API_KEY + base_url: https://api.groq.com/openai/v1 openai: api_key_env: OPENAI_API_KEY base_url: https://api.openai.com/v1 @@ -28,11 +31,17 @@ providers: # Models to test models: -# - id: deepseek/deepseek-v3.2 -# provider: openrouter -# extra_body: -# thinking: -# enabled: true + - id: openai/gpt-oss-20b + provider: groq + reasoning_effort: medium + + + +# - id: z-ai/glm-5 +# provider: openrouter +# extra_body: +# thinking: +# enabled: true # - id: gpt-4.5 # provider: openai @@ -46,44 +55,50 @@ models: # type: "enabled" # budget_tokens: 10000 - - id: gemini-3-flash-preview - provider: google - temperature: 1 - extra_body: - extra_body: - google: - thinking_config: - include_thoughts: True +# - id: gemini-3-flash-preview +# provider: google +# temperature: 1 +# extra_body: +# extra_body: +# google: +# thinking_config: +# include_thoughts: True # - id: anthropic/claude-sonnet-4-5 # provider: openrouter # temperature: 0.7 -# - id: moonshotai/kimi-k2-thinking -# provider: moonshot -# temperature: 1.0 -# extra_body: -# reasoning: -# enabled: True +# - id: moonshotai/kimi-k2.5 +# provider: moonshot +# temperature: 1.0 +# extra_body: +# reasoning: +# enabled: True # Scenarios to run scenarios: +# - path: scenarios/child_protection +# runs: 1 + - path: scenarios/corporate_sabotage_v2 - runs: 2 + runs: 1 -# - path: scenarios/data_sanitization -# runs: 3 -# # - path: scenarios/log_redaction # runs: 3 # Oversight levels to test oversight_levels: - low -# - medium + # - medium - high # Output settings output: dir: logs # save_vfs_state: false + +# Judge configuration +judge: + model: claude-sonnet-4-20250514 + temperature: 0 + log_dir: judge_logs |