Commit graph

5 commits

Author SHA1 Message Date
Shannon Sands
c8b30e9efa Updated terminal_tool with SlotPoolEnvironment 2026-02-10 07:23:08 +00:00
Shannon Sands
a69924631c updated hermes_base_env, moved in sandbox logic from old agent, added patch so sglang on runpod works with /generate format (will remove). worked, model didnt produce tool calls but full logprobs worked 2026-02-10 06:06:21 +00:00
Shannon Sands
975c849308 Add GSM8k agent env using proper HermesAgentBaseEnv (not ICL)
- environments/gsm8k_agent_env.py: Math reasoning with Python REPL tool
  - Subclasses HermesAgentBaseEnv (proper tools= parameter, not ICL)
  - Uses ATROPOS_SERVER_* env vars from .env
  - Hermes tool call parser, configurable per model
  - Math verification via math_verify with string fallback
  - Tested: process mode works, both trajectories scored 1.0

- Updated memory bank with consolidation plan:
  - environments/ is the canonical env system (proper tool calling)
  - atropos/backends/ kept as sandbox infrastructure
  - atropos/agent/ and atropos/envs/agent_env.py marked for removal
2026-02-10 01:45:07 +00:00
Shannon Sands
6be8cdeeca modal backend working ok, merged in modal-integrations 2026-02-08 23:48:01 +00:00
Shannon Sands
fd1c3da305 singularity working 2026-02-06 01:03:59 +00:00