feat: add --base-url and --api-key to generate_kernel_and_verify.py

docs/source/operation-guide/llm-track/commands.md

@@ -76,6 +76,18 @@ python scripts/generate_kernel_and_verify.py \
     --model-name claude-opus-4-6
 ```
 
+### Third-Party Providers
+
+Use `--base-url` to connect to any OpenAI-compatible provider.
+
+```bash
+python scripts/generate_kernel_and_verify.py \
+    --server-type openai \
+    --model-name <model-name> \
+    --base-url <api-endpoint> \
+    --api-key <your-api-key>
+```
+
 ## Advanced Options
 
 ### Enable Reflection

docs/source/operation-guide/llm-track/parameters.md

@@ -15,6 +15,8 @@ LLM Track command-line parameters.
 |-----------|---------|-------------|
 | `--op-name` | All | Test a single operator (e.g., `aten::add`) |
 | `--single-test` | Off | Randomly select 1 operator for quick testing |
+| `--base-url` | `http://localhost:8000/v1` | API base URL for OpenAI-compatible providers (e.g., DashScope, vLLM server) |
+| `--api-key` | Env var | API key (overrides `OPENAI_API_KEY` / `ANTHROPIC_API_KEY` env var) |
 | `--dataset` | Auto | Dataset: `KernelGenBench`, `KernelGenBench-aten`, `KernelGenBench-vllm`, `KernelGenBench-cublas` |
 | `--max-rounds` | 10 | Number of Pass@K rounds |
 | `--device-count` | 8 | Number of GPUs for verification |
@@ -63,6 +65,24 @@ Number of independent kernel samples to generate:
 - Higher values → better Pass@K coverage
 - Higher cost → more API calls
 
+### --base-url
+
+Specify a custom API endpoint for OpenAI-compatible providers:
+
+```bash
+--server-type openai --model-name <model> --base-url <endpoint>
+```
+
+### --api-key
+
+Override the default API key from environment variables:
+
+```bash
+--api-key <your-key>
+```
+
+If not set, reads from `OPENAI_API_KEY` or `ANTHROPIC_API_KEY` depending on `--server-type`.
+
 ## Output
 
 Results saved to `output/pass_at_k/<timestamp>/`:

scripts/generate_kernel_and_verify.py

@@ -740,6 +740,8 @@ def main():
     # Generation config
     parser.add_argument("--server-type", type=str, default="openai")
     parser.add_argument("--model-name", type=str, default="gpt-4o-mini")
+    parser.add_argument("--base-url", type=str, default=None, help="API base URL (for OpenAI-compatible providers)")
+    parser.add_argument("--api-key", type=str, default=None, help="API key (overrides OPENAI_API_KEY / ANTHROPIC_API_KEY env var)")
     parser.add_argument("--temperature", type=float, default=0.8)
     parser.add_argument("--max-tokens", type=int, default=16384)
     parser.add_argument("--num-workers", type=int, default=150)
@@ -793,10 +795,18 @@ def main():
     run_name = output_dir.name
 
     # Create generation config
+    # Set API key in env if provided
+    if args.api_key:
+        os.environ["OPENAI_API_KEY"] = args.api_key
+        os.environ["ANTHROPIC_API_KEY"] = args.api_key
+
+    base_url = args.base_url if args.base_url else "http://localhost:8000/v1"
+
     gen_config = GenerationConfig(
         run_name="",
         server_type=args.server_type,
         model_name=args.model_name,
+        base_url=base_url,
         temperature=args.temperature,
         max_tokens=args.max_tokens,
         num_workers=args.num_workers,