Revert "Allow benchmark runner config: openxla for inference runs. (p…

…ytorch#5939)"

This reverts commit dcdd66e.

benchmarks/README.md

@@ -66,10 +66,9 @@ python xla/benchmarks/experiment_runner.py                   \
 You can change the flags to add the configurations you are interested in. The
 `experiment_runner.py` will expand the options to all supported configurations.
 For example, in the case above, it will consider all the possible combinations
-among the flags `--dynamo`, `--xla`, and `--test`, 5 of which are supported:
+among the flags `--dynamo`, `--xla`, and `--test`, 4 of which are supported:
 
   - `dynamo=openxla_eval`, `xla=PJRT`, `test=eval`
-  - `dynamo=openxla`, `xla=PJRT`, `test=eval`
   - `dynamo=openxla`, `xla=PJRT`, `test=train`
   - `dynamo=inductor`, `xla=None`, `test=eval`
   - `dynamo=inductor`, `xla=None`, `test=train`

benchmarks/benchmark_experiment.py

@@ -88,7 +88,8 @@ def is_available(self, experiment_config):
     if experiment_config["dynamo"] == "openxla_eval" and not (
         experiment_config["xla"] and experiment_config["test"] == "eval"):
       return False
-    if experiment_config["dynamo"] == "openxla" and not experiment_config["xla"]:
+    if experiment_config["dynamo"] == "openxla" and not (
+        experiment_config["xla"] and experiment_config["test"] == "train"):
       return False
     if (experiment_config["xla"] and
         not is_xla_device_available(experiment_config["accelerator"].upper())):

benchmarks/experiment_runner.py

@@ -531,7 +531,7 @@ def parse_args(args=None):
       "--collect-full-output",
       action="store_true",
       help="""Whether to collect full output for training. Set this to true if we
-        want to verify the numerical correctness of gradients. But that may
+        want to verify the numerical correctness of graidents. But that may
         cause time measurement not accurate""",
   )