Promote SWE-Agent V2 example

docs/user-guide/agentic-chat-template.md

@@ -45,7 +45,7 @@ ROLLOUT_ARGS+=(
 
 ## Example
 
-A full multi-turn agentic setup on the session-server TITO path lives in [`examples/experimental/swe-agent-v2`](https://github.com/radixark/miles/tree/main/examples/experimental/swe-agent-v2): its launchers wire `--use-session-server` + `--tito-model glm47` + `--tito-allowed-append-roles user tool` against a real SWE agent.
+A full multi-turn agentic setup on the session-server TITO path lives in [`examples/swe-agent-v2`](https://github.com/radixark/miles/tree/main/examples/swe-agent-v2): its launchers wire `--use-session-server` + `--tito-model glm47` + `--tito-allowed-append-roles user tool` against a real SWE agent.
 
 ## Add a new model
 

docs/user-guide/rollout-endpoints.md

@@ -33,7 +33,7 @@ Key modules:
 |---|---|
 | `miles/rollout/base_types.py` | `GenerateFnInput` / `GenerateFnOutput` |
 | `miles/rollout/inference_rollout/inference_rollout_common.py` | Builds a `GenerateState` and calls the generate function |
-| `MILES_EXPERIMENTAL_ROLLOUT_REFACTOR=1` | Enables the new path (see `examples/experimental/swe-agent-v2`) |
+| `MILES_EXPERIMENTAL_ROLLOUT_REFACTOR=1` | Enables the new path (see `examples/swe-agent-v2`) |
 
 ### Generate function basics
 
@@ -162,7 +162,7 @@ Generator entry point:
 
 Example:
 
-- [`examples/experimental/swe-agent-v2`](https://github.com/radixark/miles/tree/main/examples/experimental/swe-agent-v2):
+- [`examples/swe-agent-v2`](https://github.com/radixark/miles/tree/main/examples/swe-agent-v2):
   multi-turn agentic SWE agent on the session-server TITO path, with ready-to-run launchers.
 
 Wire-up (as used by swe-agent-v2):
@@ -206,7 +206,7 @@ The hook is **entirely optional and safe to omit**:
 - It only fires when `--custom-agent-function-path` is set, so non-agentic runs
   never invoke it.
 
-See [`swe_agent_function.abort`](https://github.com/radixark/miles/blob/main/examples/experimental/swe-agent-v2/swe_agent_function.py)
+See [`swe_agent_function.abort`](https://github.com/radixark/miles/blob/main/examples/swe-agent-v2/swe_agent_function.py)
 for a reference implementation that flushes the Harbor agent server.
 
 ### Customizing the wrapper

examples/README.md

@@ -18,6 +18,7 @@ These examples provide concrete examples to leverage miles in your own RL workfl
 - **[retool](./retool)**: Demonstrates the retool functionality for tool-enabled language model generation.
 - **[search-r1](./search-r1)**: A minimal reproduction of Search-R1, featuring multi-turn conversation and tool-calling.
 - **[strands-agents](./strands-agents)**: Integration example with the Strands-Agents scaffolding framework.
+- **[swe-agent-v2](./swe-agent-v2)**: Trains coding and terminal agents with Harbor-managed sandboxes and verifier rewards.
 - **[tau-bench](./tau-bench)**: Training in an agentic multi-turn tool use environment (Tau-bench).
 - **[train_infer_mismatch_helper](./train_infer_mismatch_helper)**: Algorithmic methods for rollout correction (e.g., TIS, MIS).
 - **[true_on_policy](./true_on_policy)**: Ensures strictly equal log probabilities between inference (SGLang) and training engines.

examples/experimental/glm47-reasoning/README.md

@@ -0,0 +1,8 @@
+# GLM-4.7 reasoning launchers
+
+Experimental synchronous and fully asynchronous launchers for training the full
+GLM-4.7 model on GSM8K. These scripts use the math reward path and do not depend
+on SWE-Agent, Harbor, or an agent server.
+
+- `run-glm47-reasoning.py`: colocated synchronous training.
+- `run-glm47-reasoning-async.py`: disaggregated fully asynchronous training.

examples/experimental/swe-agent-v2/run-glm47-reasoning-async.py → examples/experimental/glm47-reasoning/run-glm47-reasoning-async.py

@@ -1,5 +1,7 @@
 """GLM-4.7 Full (355B-A32B) fully-async reasoning training with GSM8K data.
 
+This launcher is independent of the SWE-agent environment example.
+
 Disaggregated fully-async variant of run-glm47-reasoning.py: training and
 rollout run on separate nodes concurrently. Uses train_async.py and the
 fully_async_rollout module so that weight updates do not block generation.

examples/experimental/swe-agent-v2/run-glm47-reasoning.py → examples/experimental/glm47-reasoning/run-glm47-reasoning.py

@@ -1,5 +1,7 @@
 """GLM-4.7 Full (355B-A32B) reasoning training with GSM8K data.
 
+This launcher is independent of the SWE-agent environment example.
+
 Debug script: uses math (GSM8K) data instead of agentic tool use to verify
 that the training pipeline produces nonzero rewards and learns successfully.
 

examples/experimental/swe-agent-v2-amd/README.md

@@ -1,8 +1,8 @@
 # Agent V2 — Qwen3 / Qwen3-Coder on ROCm/AMD
 
-ROCm/AMD-oriented variant of [`../swe-agent-v2`](../swe-agent-v2). It is **self-contained** —
+ROCm/AMD-oriented variant of [`../../swe-agent-v2`](../../swe-agent-v2). It is **self-contained** —
 it carries its own copies of `run.py`, `generate.py`, and `swe_agent_function.py` so it runs
-without modifying the shared example. See [`../swe-agent-v2/README.md`](../swe-agent-v2/README.md)
+without modifying the shared example. See [`../../swe-agent-v2/README.md`](../../swe-agent-v2/README.md)
 for the full pipeline/architecture; this doc covers only the Qwen3 launcher and the ROCm notes.
 
 ## Qwen3 / Qwen3-Coder