-
Notifications
You must be signed in to change notification settings - Fork 75.2k
Issues
is:issue state:open
is:issue state:open
Issue creation is restricted in this repository
Search results
tf2xla: many ops fail
jit_compile=Truewith "... must be a compile-time constant" when a shape/size argument is data-dependent2.21.0tensorflow 2.21.0tensorflow 2.21.0comp:xlaXLAXLAtype:bugBugBugStatus: Open.#122057 In tensorflow/tensorflow;tf2xla:
tf.experimental.numpy.crossfails XLA compile: "Cross-products are only defined for 3-element vectors"2.21.0tensorflow 2.21.0tensorflow 2.21.0comp:xlaXLAXLAtype:bugBugBugStatus: Open.#122056 In tensorflow/tensorflow;tf2xla:
Invalid shape broadcast from f64[<=96] to [44](bounded dynamic dim) underjit_compile=True2.21.0tensorflow 2.21.0tensorflow 2.21.0comp:xlaXLAXLAtype:bugBugBugStatus: Open.#122055 In tensorflow/tensorflow;tf2xla:
-3 is out of range [0 .. 5)underjit_compile=True(eager succeeds)2.21.0tensorflow 2.21.0tensorflow 2.21.0comp:xlaXLAXLAtype:bugBugBugStatus: Open.#122054 In tensorflow/tensorflow;tf2xla:
tf.dynamic_stitchfails XLA compile with "not all indices are used" (eager succeeds)2.21.0tensorflow 2.21.0tensorflow 2.21.0comp:xlaXLAXLAtype:bugBugBugStatus: Open.#122053 In tensorflow/tensorflow;tf2xla:
Incompatible shapes: [4] vs. [8]underjit_compile=True(eager succeeds)2.21.0tensorflow 2.21.0tensorflow 2.21.0comp:xlaXLAXLAtype:bugBugBugStatus: Open.#122052 In tensorflow/tensorflow;tf2xla:
UnimplementedError: Complex types not supportedaftertf.bitcastto complex underjit_compile=True2.21.0tensorflow 2.21.0tensorflow 2.21.0comp:xlaXLAXLAtype:bugBugBugStatus: Open.#122051 In tensorflow/tensorflow;tf2xla constant folding: "AnalyzeConstantValueFallback can't handle opcode: tan"
2.21.0tensorflow 2.21.0tensorflow 2.21.0comp:xlaXLAXLAtype:bugBugBugStatus: Open.#122050 In tensorflow/tensorflow;tf2xla:
tf.sequence_maskwith a float16 maxlen fails XLA compile: "Invalid type for Range half"2.21.0tensorflow 2.21.0tensorflow 2.21.0comp:xlaXLAXLAtype:bugBugBugStatus: Open.#122049 In tensorflow/tensorflow;tf2xla:
tf.math.round/RoundToEvenfails XLA compile on integer input: "Operands to RoundToEven must be real-valued floating-point, but got S32"2.21.0tensorflow 2.21.0tensorflow 2.21.0comp:xlaXLAXLAtype:bugBugBugStatus: Open.#122048 In tensorflow/tensorflow;XLA: Igamma / Igammac / Zeta reject **identical** operand shapes — "must have equal shapes and types; got X and X"
2.21.0tensorflow 2.21.0tensorflow 2.21.0comp:opsOPs related issuesOPs related issuescomp:xlaXLAXLAtype:bugBugBugStatus: Open.#122047 In tensorflow/tensorflow;tf2xla: No registered OpKernel for XLA_CPU_JIT for
tf.nn.isotonic_regressionandtf.ragged.range2.21.0tensorflow 2.21.0tensorflow 2.21.0comp:xlaXLAXLAtype:bugBugBugStatus: Open.#122046 In tensorflow/tensorflow;