TOSA specification v0.90.0 Release Notes
Add shape operators to track shape operations for future shape inference
Continued pseudocode cleanup
Conformance testing improvements
Specify accumulator type for REDUCE_SUM
Remove redundant out_shape argument from TRANSPOSE_CONV2D
- 526f6c7 Add section of shape operators by Eric Kunze · 5 months ago v0.90.0
- 75920b1 pseudocode: fix name mismatches with operator arguments in the XML by Kevin Petit · 4 months ago
- 34f67da IDENTITY: Add missing types by Dominic Symes · 6 months ago
- 9151dd5 REDUCE_SUM: Specify the accumulator type by Dominic Symes · 7 months ago
- f43fdce Fix description of CONST output by Eric Kunze · 7 months ago
- 946ba35 Update CustomOp's attribute names by Jerry Ge · 7 months ago
- 8033f12 Add description of bias broadcast by Eric Kunze · 7 months ago
- def882e TRANSPOSE_CONV2D: Remove redundant out_shape argument by Dominic Symes · 8 months ago
- 629baef Update version to 0.90 draft by Eric Kunze · 9 months ago
- 8bf6e71 Main Conformance: Add local_bound to RFFT2D by Dominic Symes · 9 months ago
- 633a3d9 Update Custom operator arguments by Eric Kunze · 11 months ago
- fb0284e Change TOSA specification to signless types by Eric Kunze · 11 months ago
- b5b0678 Main conformance: Add local bound flag for convolutions by Dominic Symes · 10 months ago
- 830b43b Add DIM operator and operations on shape_t values by Dominic Symes · 1 year, 1 month ago
- e1f517c Increase MAX_SCALE for the 8K level by Eric Kunze · 10 months ago
- 0fc278b Add StatefulOps to TOSA specification by Jerry Ge · 10 months ago
- 12ab5da Allow bias tensors to be broadcast by Dominic Symes · 10 months ago
- 7b0f1c9 COND_IF/WHILE_LOOP: Add nesting level limit by Dominic Symes · 11 months ago
- 67f6b26 Change max rank for reduction operators to MAX_RANK by Eric Kunze · 11 months ago
- 757b0af Add Int4 to ConstOp supported data types by Jerry Ge · 12 months ago
- 14aa620 FULLY_CONNECTED: Change weights and bias to be inputs by Dominic Symes · 12 months ago
- 8e14dcd Align RESCALE zp description with pseudocode by Eric Kunze · 12 months ago
- f0bc550 Fix ErfOp to match with the rest of activation operators by Jerry Ge · 12 months ago
- 52c89f4 Add ERF activation function by Eric Kunze · 1 year, 1 month ago
- 5a20033 Update version to 0.80.0 draft by Eric Kunze · 1 year ago v0.80a0
- 57b171f Remove draft tag from 0.70 specification by Eric Kunze · 1 year ago v0.70 v0.70.0
- 5333c25 Formalise the description of operator argument types by Kevin Petit · 1 year, 1 month ago
- 98b3e33 Replace uint6_t with int8_t by Kevin Petit · 1 year, 1 month ago
- d728218 Use index_t for SLICE start/size by Kevin Petit · 1 year, 1 month ago
- c0e467a Swap order of condition and input_list arguments on COND_IF by Kevin Petit · 1 year, 1 month ago
- 42085e3 Add TOSA rank requirements to TOSA XML by Eric Kunze · 1 year, 5 months ago
- 53e7659 Fix resize_t enum name by Kevin Petit · 1 year, 1 month ago
- bc72ba8 Move the tensor size limit into the level range definition by Dominic Symes · 1 year, 1 month ago
- 544227e Add enumerations to TOSA specification by Eric Kunze · 1 year, 4 months ago
- 9bb5f5d Add missing MAX_RANK to NEGATE by Jeremy Johnson · 1 year, 2 months ago
- f28bd0a Update Argmax's rank maximum to MAX_RANK by Jerry Ge · 1 year, 2 months ago
- 6577bcd Fix typo in TILE attributes by Eric Kunze · 1 year, 2 months ago
- 1854892 SCATTER/GATHER: Replace value_t with in_out_t by Kevin Petit · 1 year, 3 months ago
- 6305460 Update version to 0.70.0 draft by Eric Kunze · 1 year, 3 months ago
- 68a3f7b Update version to 0.60.0 for release by Eric Kunze · 1 year, 3 months ago v0.60 v0.60.0
- 2fc810e Add int48 to the supported types for CONST by Eric Kunze · 1 year, 5 months ago
- 580fdd1 Fix rank limits for various data layout operations by Luke Hutton · 1 year, 4 months ago
- 87bf309 Add accumulator size attribute to AVG_POOL2D by Eric Kunze · 1 year, 5 months ago
- 13ede1d RFFT2D: Adjust output height to be same as the input by Luke Hutton · 1 year, 5 months ago
- a269fd4 SCATTER/GATHER: Remove redundant definition of index_t by Kevin Petit · 1 year, 6 months ago
- b7293ae SELECT: Replace use of cmp_t with bool_t by Kevin Petit · 1 year, 6 months ago
- 36f2d64 Use XML as source for specification version by Kevin Petit · 1 year, 6 months ago
- 4370b93 Clarify the type of condition tensors by Dominic Symes · 1 year, 6 months ago
- e4d6a1b Add Levels defintion by Dominic Symes · 1 year, 7 months ago
- 2ceeaa2 Change RESIZE border attribute to int16_t by Eric Kunze · 1 year, 8 months ago
- 3c5f0e1 Add CAST between floating point types by Eric Kunze · 1 year, 8 months ago
- ccde4d7 RESIZE: Clarify floating point types by Dominic Symes · 1 year, 8 months ago
- 8783adb RESCALE and POW: remove rank restriction by Dominic Symes · 1 year, 10 months ago
- 74a37ca Fix cut and paste error for MATMUL by Eric Kunze · 1 year, 9 months ago
- 52dd6bb Change convolution weights to always dynamic by Eric Kunze · 1 year, 10 months ago
- 58098a7 Machine parsable specification by Eric Kunze · 1 year, 10 months ago