Delete utils/matrix_logic/generate_sweep_configs.py; use infx.matrix.generate

Workflows (e2e-tests, profile), OperatorX's test-list enumerator, tests
and docs run python -m infx.matrix.generate. Historical append-only
planning still runs a base revision's own generator: -m
infx.matrix.generate from the snapshot root when the revision has the
module, else its legacy utils/matrix_logic script (older revisions).
The now-empty utils/matrix_logic directory (pytest.ini only) is removed.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

Author: functionstackx

.claude/commands/add-model-hardware.md

@@ -147,7 +147,7 @@ new entry is **required** for CI to run your config.
 python3 -c "import yaml; yaml.safe_load(open('benchmarks/single_node/srt-slurm-recipes/<recipe>'))"
 python3 -c "import yaml; yaml.safe_load(open('configs/<nvidia|amd>-master.yaml')); yaml.safe_load(open('perf-changelog.yaml'))"
 uv run --no-project --exclude-newer PT12H --python 3.12 --with pydantic --with pyyaml \
-  utils/matrix_logic/generate_sweep_configs.py test-config \
+  python -m infx.matrix.generate test-config \
   --config-files configs/<nvidia|amd>-master.yaml --config-keys <key>
 ```
 Sanity-check the generated matrix: expected layouts/concurrencies, `max-model-len` = scenario

.github/AGENT_OPERATIONS.md

@@ -80,7 +80,7 @@ The dispatch POST returns no body or run ID.
 
 Full details live in `utils/evals/EVALS.md`.
 
-`mark_eval_entries()` in `utils/matrix_logic/generate_sweep_configs.py` selects evals, which default to the 8k1k subset and run separately from throughput with `EVAL_ONLY=true`.
+`mark_eval_entries()` in `infx/matrix/generate.py` selects evals, which default to the 8k1k subset and run separately from throughput with `EVAL_ONLY=true`.
 
 - `--no-evals`: skip evals.
 - `--evals-only`: run the default selected eval subset and suppress throughput.

.github/workflows/README.md

@@ -309,7 +309,7 @@ configs/*.yaml
         │
         ▼
 ┌─────────────────────────┐
-│  generate_sweep_configs │  ← Matrix generation
+│  infx.matrix.generate   │  ← Matrix generation
 └─────────────────────────┘
         │
         ▼

.github/workflows/claude.yml

@@ -113,7 +113,7 @@ jobs:
             )
             ```
 
-            The `generate-cli-command` input accepts arguments for `generate_sweep_configs.py`. Usage: `generate_sweep_configs.py` `[-h]` `{full-sweep,test-config}`
+            The `generate-cli-command` input accepts arguments for `python -m infx.matrix.generate`. Usage: `python -m infx.matrix.generate` `[-h]` `{full-sweep,test-config}`
 
             **Subcommand reference:**
             - `full-sweep`: Use this subcommand with filter flags like `--model-prefix`, `--framework`, `--precision`, `--runner-type`, `--min-conc`, `--max-conc`, `--seq-lens`. This is the primary subcommand for running benchmarks.

.github/workflows/e2e-tests.yml

@@ -314,9 +314,9 @@ jobs:
                     if [ "$EVALS_ONLY" = "true" ]; then
                       GENERATE_ARGS+=(--evals-only)
                     fi
-                    CONFIG_JSON=$(uv run --no-project --exclude-newer PT12H --python 3.12 --with pydantic --with pyyaml \
-                      python "${GITHUB_WORKSPACE}/utils/matrix_logic/generate_sweep_configs.py" \
-                      "${GENERATE_ARGS[@]}")
+                    CONFIG_JSON=$(env PYTHONPATH="${GITHUB_WORKSPACE}" \
+                      uv run --no-project --exclude-newer PT12H --python 3.12 --with pydantic --with pyyaml \
+                      python -P -m infx.matrix.generate "${GENERATE_ARGS[@]}")
                   fi
                   CONFIG_JSON=$(printf '%s' "$CONFIG_JSON" | env PYTHONPATH="$PRIORITY_ROOT" \
                     uv run --no-project --exclude-newer PT12H --python 3.12 --with pydantic --with pyyaml \

.github/workflows/profile.yml

@@ -85,8 +85,9 @@ jobs:
           fi
           source "$PRIORITY_ROOT/benchmarks/benchmark_lib.sh" --validation-only
           check_env_vars INPUTS_CONFIG_FILE INPUTS_CONFIG_KEY INPUTS_CONC PR_LABELS
-          CONFIG_JSON=$(uv run --no-project --exclude-newer PT12H --python 3.12 --with pydantic --with pyyaml \
-            python "${GITHUB_WORKSPACE}/utils/matrix_logic/generate_sweep_configs.py" \
+          CONFIG_JSON=$(env PYTHONPATH="${GITHUB_WORKSPACE}" \
+            uv run --no-project --exclude-newer PT12H --python 3.12 --with pydantic --with pyyaml \
+            python -P -m infx.matrix.generate \
             test-config --config-files "$INPUTS_CONFIG_FILE" --config-keys "$INPUTS_CONFIG_KEY" --conc "$INPUTS_CONC")
           CONFIG_JSON=$(printf '%s' "$CONFIG_JSON" | env PYTHONPATH="$PRIORITY_ROOT" uv run --no-project --exclude-newer PT12H --python 3.12 --with pyyaml \
             python -P "${PRIORITY_ROOT}/utils/ci_priority.py" \

benchmarks/multi_node/srt-slurm-recipes/RECIPES.md

@@ -59,7 +59,7 @@ srtctl migrate --verify -f benchmarks/multi_node/srt-slurm-recipes/dsr1/sglang
 srtctl migrate --in-place -f benchmarks/multi_node/srt-slurm-recipes/dsr1/sglang
 # Repeat for the other model/engine directories.
 # Use the pinned TileRT fork for glm5.1/tilert/.
-python -m pytest utils/matrix_logic/ -q
+python -m pytest infx/tests/matrix/ -q
 python -m infx.matrix.generate full-sweep \
   --config-files configs/nvidia-master.yaml \
   --framework dynamo-sglang dynamo-trt dynamo-vllm --multi-node

benchmarks/multi_node/srt-slurm-recipes/RECIPES_zh.md

@@ -59,7 +59,7 @@ srtctl migrate --verify -f benchmarks/multi_node/srt-slurm-recipes/dsr1/sglang
 srtctl migrate --in-place -f benchmarks/multi_node/srt-slurm-recipes/dsr1/sglang
 # 对其他模型/引擎目录重复执行。
 # 迁移 glm5.1/tilert/ 时，使用固定提交的 TileRT 分支仓库。
-python -m pytest utils/matrix_logic/ -q
+python -m pytest infx/tests/matrix/ -q
 python -m infx.matrix.generate full-sweep \
   --config-files configs/nvidia-master.yaml \
   --framework dynamo-sglang dynamo-trt dynamo-vllm --multi-node

docs/MODELS.md

@@ -175,7 +175,7 @@ Other offloading tiers, including NVMe KV cache offloading, are outside the init
 
 ## Notes
 
-- The `Prefix` column is the canonical `model-prefix` used in `configs/*-master.yaml` and by `generate_sweep_configs.py --model-prefix`.
+- The `Prefix` column is the canonical `model-prefix` used in `configs/*-master.yaml` and by `python -m infx.matrix.generate --model-prefix`.
 - "Retired" means the model no longer has any active scenario. Retired models' configs are deleted from the master configs; the former `configs/deprecated/` archive was removed in #3464, so past settings live in git history and `perf-changelog.yaml`.
 - Deprecating a precision (e.g. Qwen3.5 bf16) or one arm of an A/B pair (e.g. non-MTP) narrows a model's recipe coverage without retiring the model. The model stays listed as active as long as one scenario still runs.
 - `dsr1` began as the DeepSeek-V3 workflow templates in the initial repo import and was switched to DeepSeek-R1 benchmarking on 2025-08-13 (renamed `dsv3` → `dsr1` on 2025-08-20).

docs/MODELS_zh.md

@@ -175,7 +175,7 @@ InferenceX 支持 SGLang 和 vLLM 双方的维护者，并响应 AI 实验室和
 
 ## 说明
 
-- 「前缀」列为 `configs/*-master.yaml` 中的规范 `model-prefix`，同时用于 `generate_sweep_configs.py --model-prefix`。
+- 「前缀」列为 `configs/*-master.yaml` 中的规范 `model-prefix`，同时用于 `python -m infx.matrix.generate --model-prefix`。
 - 「退役」指该模型已无任何启用场景。退役模型的配置直接从主配置中删除；原 `configs/deprecated/` 归档目录已在 #3464 中删除，历史设置保留在 Git 历史与 `perf-changelog.yaml` 中。
 - 弃用某一精度（如 Qwen3.5 bf16）或 A/B 对照中的某一分支（如非 MTP），只是收窄该模型的配方覆盖范围，并不等于模型退役；只要仍有一个场景在运行，该模型即继续列为启用状态。
 - `dsr1` 最初以 DeepSeek-V3 workflow 模板的形式随仓库首次导入，2025-08-13 切换为 DeepSeek-R1 基准测试（2025-08-20 将 `dsv3` 重命名为 `dsr1`）。