emx-onnx-cgen 0.3.8__py3-none-any.whl → 0.4.1.dev0__py3-none-any.whl

{emx_onnx_cgen-0.3.8.dist-info → emx_onnx_cgen-0.4.1.dev0.dist-info}/METADATA

@@ -1,7 +1,10 @@
 Metadata-Version: 2.4
 Name: emx-onnx-cgen
-Version: 0.3.8
+Version: 0.4.1.dev0
 Summary: emmtrix ONNX-to-C Code Generator
+Project-URL: Homepage, https://github.com/emmtrix/emx-onnx-cgen
+Project-URL: Repository, https://github.com/emmtrix/emx-onnx-cgen
+Project-URL: Issues, https://github.com/emmtrix/emx-onnx-cgen/issues
 Requires-Python: >=3.10
 Description-Content-Type: text/markdown
 
@@ -39,10 +42,10 @@ Key characteristics:
 
 - CLI for ONNX-to-C compilation and verification.
 - Deterministic codegen with explicit tensor shapes and loop nests.
-- Minimal C runtime templates in `templates/`.
+- Minimal C runtime templates in `src/emx_onnx_cgen/templates/`.
 - ONNX Runtime comparison for end-to-end validation.
 - Official ONNX operator coverage tracking.
-- Support for a wide range of ONNX operators (see `OFFICIAL_ONNX_FILE_SUPPORT.md`).
+- Support for a wide range of ONNX operators (see `ONNX_SUPPORT.md`).
 - Supported data types:
   - `float`, `double`, `float16`
   - `int8_t`, `uint8_t`, `int16_t`, `uint16_t`, `int32_t`, `uint32_t`, `int64_t`, `uint64_t`
@@ -89,13 +92,16 @@ emx-onnx-cgen compile <model.onnx> <output.c> [options]
 
 Options:
 
-- `--template-dir`: Directory containing the C templates (default: `templates`).
+- `--model-base-dir`, `-B`: Base directory for resolving the model path (example: `emx-onnx-cgen compile --model-base-dir /data model.onnx out.c`).
+- `--color`: Colorize CLI output (`auto`, `always`, `never`; default: `auto`).
 - `--model-name`: Override the generated model name (default: output file stem).
 - `--emit-testbench`: Emit a JSON-producing `main()` testbench for validation.
 - `--emit-data-file`: Emit constant data arrays into a companion `_data` C file.
-- `--large-weight-threshold`: Store weights larger than this element count in a binary file (default: `1048576`; set to `0` to disable).
-- `--large-temp-threshold-bytes`: Mark temporary buffers larger than this threshold as static (default: `1024`).
+- `--large-weight-threshold`: Store weights in a binary file once the cumulative byte size exceeds this threshold (default: `102400`; set to `0` to disable).
+- `--large-temp-threshold`: Mark temporary buffers larger than this threshold as static (default: `1024`).
 - `--no-restrict-arrays`: Disable `restrict` qualifiers on generated array parameters.
+- `--fp32-accumulation-strategy`: Accumulation strategy for float32 inputs (`simple` uses float32, `fp64` uses double; default: `fp64`).
+- `--fp16-accumulation-strategy`: Accumulation strategy for float16 inputs (`simple` uses float16, `fp32` uses float; default: `fp32`).
 
 ### `verify`
 
@@ -105,13 +111,20 @@ emx-onnx-cgen verify <model.onnx> [options]
 
 Options:
 
-- `--template-dir`: Directory containing the C templates (default: `templates`).
+- `--model-base-dir`, `-B`: Base directory for resolving the model and test data paths (example: `emx-onnx-cgen verify --model-base-dir /data model.onnx --test-data-dir inputs`).
+- `--color`: Colorize CLI output (`auto`, `always`, `never`; default: `auto`).
 - `--model-name`: Override the generated model name (default: model file stem).
 - `--cc`: Explicit C compiler command for building the testbench binary.
-- `--large-weight-threshold`: Store weights larger than this element count in a binary file (default: `1024`).
-- `--large-temp-threshold-bytes`: Mark temporary buffers larger than this threshold as static (default: `1024`).
+- `--large-weight-threshold`: Store weights in a binary file once the cumulative byte size exceeds this threshold (default: `102400`).
+- `--large-temp-threshold`: Mark temporary buffers larger than this threshold as static (default: `1024`).
 - `--max-ulp`: Maximum allowed ULP distance for floating outputs (default: `100`).
-- `--runtime`: Runtime backend for verification (`onnxruntime` or `onnx-reference`, default: `onnx-reference`).
+- `--atol-eps`: Absolute tolerance as a multiple of machine epsilon for floating outputs (default: `1.0`).
+- `--runtime`: Runtime backend for verification (`onnxruntime` or `onnx-reference`, default: `onnxruntime`).
+- `--temp-dir-root`: Root directory in which to create a temporary verification directory (default: system temp dir).
+- `--temp-dir`: Exact directory to use for temporary verification files (default: create a temporary directory).
+- `--keep-temp-dir`: Keep the temporary verification directory instead of deleting it.
+- `--fp32-accumulation-strategy`: Accumulation strategy for float32 inputs (`simple` uses float32, `fp64` uses double; default: `fp64`).
+- `--fp16-accumulation-strategy`: Accumulation strategy for float16 inputs (`simple` uses float16, `fp32` uses float; default: `fp32`).
 
 How verification works:
 
@@ -122,9 +135,14 @@ How verification works:
    directory.
 3. **Run runtime backend**: the JSON inputs from the testbench are fed to the
    selected runtime (`onnxruntime` or `onnx-reference`) using the same model.
-4. **Compare outputs**: floating outputs are compared by maximum ULP distance
-   (see https://www.emmtrix.com/wiki/ULP_Difference_of_Float_Numbers for the
-   ULP definition and algorithm); non-floating outputs must match exactly.
+   The compiler no longer ships a Python runtime evaluator.
+4. **Compare outputs**: floating outputs are compared by maximum ULP distance.
+   Floating-point verification first ignores very small differences up to
+   **--atol-eps × [machine epsilon](https://en.wikipedia.org/wiki/Machine_epsilon) of
+   the evaluated floating-point type**, treating such values as equal. For
+   values with a larger absolute difference, the ULP distance is computed, and
+   the maximum ULP distance is reported; non-floating outputs must match
+   exactly.
    Missing outputs or mismatches are treated as failures.
 5. **ORT unsupported models**: when using `onnxruntime`, if ORT reports
    `NOT_IMPLEMENTED`, verification is skipped with a warning (exit code 0).
@@ -147,8 +165,8 @@ runtime.
 
 ## Official ONNX test coverage
 
-See [`OFFICIAL_ONNX_FILE_SUPPORT.md`](OFFICIAL_ONNX_FILE_SUPPORT.md) for the generated support matrix.
-See [`SUPPORT_OPS.md`](SUPPORT_OPS.md) for operator-level support derived from the expectation JSON files.
+See [`ONNX_SUPPORT.md`](https://github.com/emmtrix/emx-onnx-cgen/blob/v0.4.0/ONNX_SUPPORT.md) for the generated support matrix.
+See [`SUPPORT_OPS.md`](https://github.com/emmtrix/emx-onnx-cgen/blob/v0.4.0/SUPPORT_OPS.md) for operator-level support derived from the expectation JSON files.
 
 ## Maintained by
 

emx_onnx_cgen-0.4.1.dev0.dist-info/RECORD

@@ -0,0 +1,190 @@
+emx_onnx_cgen/__init__.py,sha256=jUSbu1kJ0krzVTYEcph3jCprBhD7tWNtiSdL6r29KrM,221
+emx_onnx_cgen/__main__.py,sha256=iC1lLVtR6-TmpL6OxXcy3oIntExUtajn9-q627R1XyI,140
+emx_onnx_cgen/_build_info.py,sha256=2a3tiwcONr275Rn1dg8CjAqlSrdAJUd_LYt3hxGyRa4,112
+emx_onnx_cgen/_version.py,sha256=ptLWt_Y6Ljpzx3CLPNc1ziUZh_FZbRUr08VOhJPCHpE,717
+emx_onnx_cgen/cli.py,sha256=YJjo4r73IcZ8fMClqmFtnmwyv1nzjPixjnBr67AJnjo,53202
+emx_onnx_cgen/compiler.py,sha256=pR9iEuQd8UQvD-rxh2O0Rte1SlnJh_ns1F1JM5JxX3Q,21832
+emx_onnx_cgen/determinism.py,sha256=6O43NfbRfLfKUQqL4cHUS-EaPhFCLGrJ4kKB5C7z5TQ,1103
+emx_onnx_cgen/dtypes.py,sha256=jRx3BBvk0qFW14bngoL1B7L_IRasyNJ4jqhpM5YhcOM,1335
+emx_onnx_cgen/errors.py,sha256=HpOv95mTgr9ZX2gYe1RtwVMbPskh7zkqjU_FgAD-uIM,363
+emx_onnx_cgen/onnx_import.py,sha256=FrxAzwtTXj-XhN5KcqqyqkN-Q_PNF9cFlU1OlAXdi7Q,29041
+emx_onnx_cgen/onnxruntime_utils.py,sha256=mEsC1x00M1jyBgVBKqnKoqx6H1tdgsFFUy7rbITs3bs,308
+emx_onnx_cgen/ops.py,sha256=zrVmDMfwZC3RnPWl_YOLUEdr3j9MC-WQHXV0dETufWc,16888
+emx_onnx_cgen/testbench.py,sha256=-NbqD1aC7OXvFMLiLzd2IPObenQdHFH85cNxNSB1GeY,640
+emx_onnx_cgen/validation.py,sha256=KFdUdGjQbzTj1szCJcjxnTi8f5l6ywNgCB9abbBpTbM,2360
+emx_onnx_cgen/verification.py,sha256=mtkUqJpJbXWHnWB-Rpt0n_cSeh4rPBufHFP7y5KH3TY,2084
+emx_onnx_cgen/codegen/__init__.py,sha256=EqCc1e0n2PBXbJKTDq6VXehBNIUVwqS0OYC8e68CYqM,444
+emx_onnx_cgen/codegen/c_emitter.py,sha256=tPut1jHbBFuZ1auRKgUvHRnGDMKNe9M8efqDT2xaR0Y,520147
+emx_onnx_cgen/codegen/emitter.py,sha256=udcsqJNr46TFHiyVv5I4wdVH8ll6Bi4VqcR1VvofbnY,92
+emx_onnx_cgen/ir/__init__.py,sha256=fD2D8qxlGoCFJb0m9v6u3XTgzSxDOhB4cfLBiCLovzg,102
+emx_onnx_cgen/ir/context.py,sha256=Xo1UCezDdGLF195GSLP4xARDEN6dBwLRez-EqUFTTmY,3813
+emx_onnx_cgen/ir/model.py,sha256=Bjy8edpmeET_U-H44g-uEe3HUumUMiZNUQbOf101MOg,1295
+emx_onnx_cgen/ir/op_base.py,sha256=RbJRg_V9Zh2a9nKPUAUvGIWv16Y_l5t1JcmifOSdHQ0,18358
+emx_onnx_cgen/ir/op_context.py,sha256=9CZCUNJLsV4cJsYmJqWbaDrwQd4sr-9Ot1PmPSqGAto,2103
+emx_onnx_cgen/ir/ops/__init__.py,sha256=kQxoj0uMI-1ngOjWHEtX_deLSwrPfwQGgMmjis7wSJA,2905
+emx_onnx_cgen/ir/ops/elementwise.py,sha256=H7MOYYNSmz8Jsy7RaufF3BA-AmGX21JEPeIsipUOu2Q,8174
+emx_onnx_cgen/ir/ops/misc.py,sha256=-nRybPqwar0uS4CiZDH8EScPyBALAFfSw9BxdJGhO-o,16234
+emx_onnx_cgen/ir/ops/nn.py,sha256=6oVU6pddR28E4Q52V5E8pAWxRV3r0dgoT_RblEd1jYA,26893
+emx_onnx_cgen/ir/ops/reduce.py,sha256=ePIJwzlswsktvJazMtOd0YuazCHS2dXUwrqCKhUauk4,2263
+emx_onnx_cgen/lowering/__init__.py,sha256=pD_983kCqKV-M6g0z62XLUAwHGExb2lY1HptTXegMbk,1717
+emx_onnx_cgen/lowering/adagrad.py,sha256=DuW3MeNNJjhXz1k7XI9JDwfgWr-TyD5Q-B9eAZrNecM,4797
+emx_onnx_cgen/lowering/arg_reduce.py,sha256=6D785Z_yn9n0uabn9W5v6-YsjqHZbpvOa8L9oySh5U0,3241
+emx_onnx_cgen/lowering/attention.py,sha256=-Il_8AQMuwQtq-2-RkVyVfnvtRJuO61Cv1PlMIypxEc,16477
+emx_onnx_cgen/lowering/average_pool.py,sha256=r-agefzjj99Fw2CP1ElhvYHiYd3R7zlPvJOfzKJcjek,12619
+emx_onnx_cgen/lowering/batch_normalization.py,sha256=_i-vwlhuAQYqxJIezHaxeqcmISV66Y_5o929_FTtMZg,3976
+emx_onnx_cgen/lowering/bernoulli.py,sha256=Fn-t0fgve_Ns2LBzilHwnLE30ONnNNxtJu7IxVcwcfw,2435
+emx_onnx_cgen/lowering/cast.py,sha256=J2Tf7MprIcZjsgVLGsaccpbyvftfXfm57o--Il-8GlQ,2841
+emx_onnx_cgen/lowering/common.py,sha256=We_bZ33DE8TW7SBG70djC1VmXjTivsB0XWqQTFEK0XI,18446
+emx_onnx_cgen/lowering/concat.py,sha256=p40vipQsbyymVqV4YO2JWVNiuqkU-xyVqqrCI8iujB4,2499
+emx_onnx_cgen/lowering/constant_of_shape.py,sha256=N01UvbVroDk08FTbBMndrLYIzI0G6M0UQuCr4oxpP40,3197
+emx_onnx_cgen/lowering/conv.py,sha256=oKJcs4foIewuTFmJ4jbzZm8mz58LyqgO3qSjc3ODx8I,7394
+emx_onnx_cgen/lowering/conv_integer.py,sha256=4TCdUDSijyXo_eEhp0sPX05G2OrmjAdCJ7yzgMp2BuQ,3814
+emx_onnx_cgen/lowering/conv_transpose.py,sha256=10K7nhQ60p0PAB3qxmeazm2tbsSS1GDeINBk7VzsH1U,11153
+emx_onnx_cgen/lowering/cumsum.py,sha256=9E0C5NtvPt6g5T4QLdIOeDkXaZNzyDklus2-qu2B7eA,4114
+emx_onnx_cgen/lowering/depth_space.py,sha256=i7INioNkofBxFlZW9y0W_qA6mp67_FAXouhKCiB9RKc,4206
+emx_onnx_cgen/lowering/dequantize_linear.py,sha256=XQ8OOIcX_8dtuS5VHWCo7FhYtIOEJeY1aMfEf5hd4Jk,4762
+emx_onnx_cgen/lowering/dropout.py,sha256=MZ4YrB-jvUFXpIKE5kOLyrEF5uy5dh0yjJH6Rj8KlMs,1764
+emx_onnx_cgen/lowering/einsum.py,sha256=MWAgWVOzP38RSOxJABwvYU6ykD9odmhrmddXinmFs7s,6117
+emx_onnx_cgen/lowering/elementwise.py,sha256=xvA1ifzLherNXbe9_EYYuyFcREk_uRcuUw6Jo38f7h8,15967
+emx_onnx_cgen/lowering/expand.py,sha256=y0h1x2xh6Oqtblm6TbELB6_I4fsquU3YuZoB4mZJeTo,525
+emx_onnx_cgen/lowering/eye_like.py,sha256=QBiHWYZbgK4uiUYWuS7WHCMBGMSG0paNZM84OYmGb7c,1723
+emx_onnx_cgen/lowering/flatten.py,sha256=6h-TQNy9iq5hfXR9h2clUrc2eHmZP9gAb9KbCSJdV20,2131
+emx_onnx_cgen/lowering/gather.py,sha256=XrGpsdOu0Rg2mKgKcGgcK4_BOjUlbBuTxFiBDhVp3lA,1054
+emx_onnx_cgen/lowering/gather_elements.py,sha256=cCp2UFOjktgEfS9s9npMS_BXklBkpMpD7UhIIMhQ-_Y,2318
+emx_onnx_cgen/lowering/gather_nd.py,sha256=rmr_ijeSeCrZ_R_QPwdoHPQUCe8nE0YRSv2NjUiiFjY,3090
+emx_onnx_cgen/lowering/gemm.py,sha256=2P9SKDGlXC8wUK363_QKS0H5bPjsuEBxPTPs89WvC5k,791
+emx_onnx_cgen/lowering/global_max_pool.py,sha256=orUQjQRK_Fm_g9ktUTYpZ74_LxnAN0tIEDaQ8OQYi1k,2058
+emx_onnx_cgen/lowering/grid_sample.py,sha256=FFbK-jrjqFLwSUu7BfSZC9So7MeCZprGKG5N4XQUxR4,5217
+emx_onnx_cgen/lowering/group_normalization.py,sha256=Ep7toUW9sHvMHb2EwNpgayygTW-TN62ooVLdaF0z9_c,2653
+emx_onnx_cgen/lowering/gru.py,sha256=xsKcYoiFBdiMI0MlHEtjD2Sw0ZIfeJEPMmaIh--3bzI,11212
+emx_onnx_cgen/lowering/hamming_window.py,sha256=v5qEaxcO45JqTPWUf5a6SKaZvfqmNUqkRQijvdZfEoA,3559
+emx_onnx_cgen/lowering/hardmax.py,sha256=VW00Te4FlmeRfD96lvbPifYU8u1DFNqClF4NsioRcBk,566
+emx_onnx_cgen/lowering/identity.py,sha256=z_iK9zTDUJ4_7OcYG2_acfj4tIXEgUG5a38p3I559uM,1961
+emx_onnx_cgen/lowering/instance_normalization.py,sha256=XrDOAo8Af7yDObtAAJ006dVCN175cWPb5Wvh61PE7xs,1939
+emx_onnx_cgen/lowering/layer_normalization.py,sha256=RjRn1sPFupB8n3RsA8O9p5vDmfmj2Q6hjMVhSFzfLkU,4518
+emx_onnx_cgen/lowering/logsoftmax.py,sha256=996JK-b_3DzYD9hywJ_ce8V8LbM1JDzzgXD0G8ONqBk,584
+emx_onnx_cgen/lowering/lp_normalization.py,sha256=il1fBWan8DwZ3dlRVSJWVhMpzHDYtwjh1YJaNm6palY,1701
+emx_onnx_cgen/lowering/lp_pool.py,sha256=ZweJqeMazliY0OvotznFf03Jsd2jOe0cA6vv_Ap0yZ4,5298
+emx_onnx_cgen/lowering/lrn.py,sha256=rJ_7ISllYphbHKmlMv3c5IwqPl-oZrEKWux7QCdjqIQ,3359
+emx_onnx_cgen/lowering/lstm.py,sha256=RVe0qGesoK-FfWeV0vCKCkoWD32Fv_C22LnQLFLr4Tc,12294
+emx_onnx_cgen/lowering/matmul.py,sha256=i4Ve0gYFhE-oKsI_bCQedbVgrh4ZbzjZrcPdTrjAcVQ,520
+emx_onnx_cgen/lowering/maxpool.py,sha256=0XoazajqrB5So-hEnR73LOSsdF7ZnguVNAc9NSjK6Q4,7483
+emx_onnx_cgen/lowering/mean_variance_normalization.py,sha256=tFeDgrocZO5Q5hNBaFl4cTFpKTPNVmRH9-FZircEffA,1864
+emx_onnx_cgen/lowering/negative_log_likelihood_loss.py,sha256=J5VfAQN2bIrt8D4_6KIGxRBk4Q9ykJvlqJftCrqy-jc,9333
+emx_onnx_cgen/lowering/non_max_suppression.py,sha256=9EeHm2aF7QBmP-s23r43VDgRvGyFWcNcI1s_jYPqln0,5749
+emx_onnx_cgen/lowering/nonzero.py,sha256=qjDlI_0s37hK-arOD-Bm_Ft9N_gTVt0X3OEqxuP1sR0,1626
+emx_onnx_cgen/lowering/one_hot.py,sha256=JGJsA35Q5hyX7nutNVJMGgTgcFxlAlolH9k4igVc2s0,4341
+emx_onnx_cgen/lowering/optional_has_element.py,sha256=Wdo-HKAJFXqBSBPfugqRy7XPvliB78IgGsazvZx1Fmk,970
+emx_onnx_cgen/lowering/pad.py,sha256=Z8361NQCwypKfTnS8-0rylX6P-S8xLU6QLbahVzxrzw,10405
+emx_onnx_cgen/lowering/qlinear_matmul.py,sha256=gsV8CAB9_PhPuCGBYEvqfhby3uHQ6-4lyfDI2Xgvw0c,7899
+emx_onnx_cgen/lowering/qlinear_mul.py,sha256=Esqx2BxJh9KXcnM0RKCagQAJSQYZ446Za0KazUHlKDw,4761
+emx_onnx_cgen/lowering/quantize_linear.py,sha256=yJOvZbGxI8HcZ_Zl9VO49qJVfZ5FwNoDq5TjTiGzKmg,4760
+emx_onnx_cgen/lowering/range.py,sha256=yaRvLHLlWNvvg-IO590jSVPv2dWrJjPWXyysSNOj0IY,3452
+emx_onnx_cgen/lowering/reduce.py,sha256=E4wuNpFyy_thi9CFwdhYRusPGk5E9cnLUqp5l_S4Rik,18209
+emx_onnx_cgen/lowering/registry.py,sha256=tNmnP6ZhIrKv83Q6VdfkTLSsw6P8cqch-nqSWpURYX8,2002
+emx_onnx_cgen/lowering/reshape.py,sha256=__sOC1FvUCIEdbHObNcL3s97bQkKVhu7ecB9M2AEIu8,13108
+emx_onnx_cgen/lowering/resize.py,sha256=XCTUppSDj9-GyztBORIuK1MJMxelA3DU_NZzfsVIlgQ,14633
+emx_onnx_cgen/lowering/rms_normalization.py,sha256=pWu5u0TqHZaL3rh07MtA6eOP0zLzNCoQ84f1V0un2Iw,2525
+emx_onnx_cgen/lowering/rotary_embedding.py,sha256=IfDxuUCJqFIK8SCviYXZfdJcrgg8tjT2ofYFUP2uv8c,6068
+emx_onnx_cgen/lowering/scatter_nd.py,sha256=WuNxsMQmCTXgqen5rygpAbZIsfca537lvvFPakn0rJU,3210
+emx_onnx_cgen/lowering/shape.py,sha256=Yy4bVL2oUI2MHMm44HbQEEiR2zK2Rezn78urFZl_M60,2536
+emx_onnx_cgen/lowering/size.py,sha256=Mfj2x0zvDrhMAcmhXI5F63dzd3w3ZT2IxfI0jMbTSuQ,1250
+emx_onnx_cgen/lowering/slice.py,sha256=LpJlCZNTlxpNGH2J88S0a54B5_BubtlOXmQ150fHwaM,15165
+emx_onnx_cgen/lowering/softmax.py,sha256=ob9xaTF5bpk43eNb3BeTLSFOWFi1JAJW65I_mIW9cZ0,566
+emx_onnx_cgen/lowering/softmax_cross_entropy_loss.py,sha256=B6h23sGBZLdpKcbtoQUhVwfLrdSJwNcbCoPoDc3rTc0,5219
+emx_onnx_cgen/lowering/split.py,sha256=zv-W5O6K6Q5reZa7UhJ53tbCFTygt1wsCiQEGD_HgSE,7098
+emx_onnx_cgen/lowering/squeeze.py,sha256=p9bER1Jkc8_6BGjsD3b7zhuak11eywoQhVFIvJ9Vzj0,6084
+emx_onnx_cgen/lowering/tensor_scatter.py,sha256=1Wqb9XsNNj1CEKnH3Vx45xh3QQbxHF9L90ycVbcsy44,4485
+emx_onnx_cgen/lowering/tfidf_vectorizer.py,sha256=DgvEtY8VM94dmSS5EjeVYnXOJ6NZbY_6nvrzphsZN0g,7189
+emx_onnx_cgen/lowering/tile.py,sha256=4kBobxBKz9OtXvTvwOw_k7_ckBFwiYZGmTi0AlAEvZI,4536
+emx_onnx_cgen/lowering/topk.py,sha256=bXwzDFoKz7C3znBlH1EgZOi62CSYX37FMN9j2F1sKgI,4656
+emx_onnx_cgen/lowering/transpose.py,sha256=EVKlbeX1MFwgq62qq8v4EG4yCAoh2RcYhggcHrdpLQU,2071
+emx_onnx_cgen/lowering/trilu.py,sha256=OjJjyo2ZRcfo9UGH8Zfq4o0PR6YDeoHSj8DzMu0w318,3266
+emx_onnx_cgen/lowering/unsqueeze.py,sha256=_2SjIqdxmR2EvFjruV86cl9aQmJcfg9IO7yolaZR2cI,5841
+emx_onnx_cgen/lowering/upsample.py,sha256=OBcJKuAZ6vGBngPlWFyyfdjK4pg-E2scydQ5KASk8XE,5285
+emx_onnx_cgen/lowering/variadic.py,sha256=KFwMG3FgCp3LGRoUTeWLNz5pIG1WHyj2Wb2w89AZ9pI,1869
+emx_onnx_cgen/lowering/where.py,sha256=SXWgJT2VA87GL8ZjvcvabhepR4DJ_mrADDq7bxJ9F5Y,2490
+emx_onnx_cgen/templates/adagrad_op.c.j2,sha256=t4cvRqoIrexiXSFLct7r_lc8mzRVUjyKBevl3Zks2Zk,905
+emx_onnx_cgen/templates/arg_reduce_op.c.j2,sha256=FJ-XKruwY6slb_hMHmMtKS3N_1YVhRGsKYQraWQemUg,753
+emx_onnx_cgen/templates/attention_op.c.j2,sha256=jIsMjyGcb996QfnAS0rP3IIgN7BkA750V8Y1yiYiRR0,9506
+emx_onnx_cgen/templates/average_pool_op.c.j2,sha256=BXr8FhcCCMdMjIfSi4kntvVLvMRq3nWOXTFIYUJJ8Bw,6480
+emx_onnx_cgen/templates/batch_norm_op.c.j2,sha256=Mxrl0Y77q9vDvCUqnKjdHSrjt9N69q14cApz9XaG8fk,566
+emx_onnx_cgen/templates/bernoulli_op.c.j2,sha256=gVrlvYzWbcA3-30sbk_Z_JlYJ4AHwVA2DJqFyaHmk8A,1117
+emx_onnx_cgen/templates/binary_op.c.j2,sha256=YGS5j48cSjMrHztVTXOtSI7EKBihfHwOt5vpDgruiAM,539
+emx_onnx_cgen/templates/cast_op.c.j2,sha256=9BaxC3x_ca-4Pi8sZ64KD5cxXy7-mS-vARkordf4Ens,419
+emx_onnx_cgen/templates/clip_op.c.j2,sha256=Qloge-NSIPQ-niIQc6SrWpJFdqFQAp1M_O5yn9VzrdU,576
+emx_onnx_cgen/templates/concat_op.c.j2,sha256=3XVNRRcoPZW34X5jecTX_1AM5nEAhCuhZG8QG03K5-0,1319
+emx_onnx_cgen/templates/constant_of_shape_op.c.j2,sha256=fl8PHmO81wOms-w3ndZ5UODMQU7yIu9jIoTZrg7cHVA,380
+emx_onnx_cgen/templates/conv_integer_op.c.j2,sha256=vydhDC2A6-RflRxvJg2f5WFuGt6cY5HDokIdgolDiYc,2439
+emx_onnx_cgen/templates/conv_op.c.j2,sha256=tvhZ1Y2z6KIgk93FAEhkgKBsBqdBST4U-cJCPdfx-Rs,2284
+emx_onnx_cgen/templates/conv_transpose_op.c.j2,sha256=nM7MY1qwgFe9kTcf47i6Kq8Nah9KIrP8S7AbDlG66pQ,2727
+emx_onnx_cgen/templates/cumsum_op.c.j2,sha256=6RXVRl5NnRr8jEV5VpuYk0E-2k6GbIwZfXpbSihV8GU,2069
+emx_onnx_cgen/templates/depth_to_space_op.c.j2,sha256=LObJ8_QCQubdut_KAozzQaVjFxuGPmr5fOOzdJOHlTc,1264
+emx_onnx_cgen/templates/dequantize_linear_op.c.j2,sha256=z0AyDco-siBR5Nm7h02u7p0avGSiTxm1c1Z3dJ2ym4o,483
+emx_onnx_cgen/templates/einsum_op.c.j2,sha256=L-mft265aXvBkfm33JSDozKDue6wK5SpbRYWT-T7Vvc,3179
+emx_onnx_cgen/templates/expand_op.c.j2,sha256=4zGcRy6E5oDQosUVkVpv_gBae-tPoz3PuDBghwxVTUY,593
+emx_onnx_cgen/templates/eye_like_op.c.j2,sha256=pFh7UgWeHKJ7dKhW-VEqLTYarjXpYSRfvCB00vmUAlc,1043
+emx_onnx_cgen/templates/gather_elements_op.c.j2,sha256=6CFztzreMZmUHXh2sr9sscPb9aBJJgBIo_XTTSgG0Xw,560
+emx_onnx_cgen/templates/gather_nd_op.c.j2,sha256=1CBnPSsUEgLrnv9MnKxT1IxwtXIO7ghOW69jq4N9UZs,1037
+emx_onnx_cgen/templates/gather_op.c.j2,sha256=76kTx8XV_iY3T5hLC0pxV7RLErvEHgcsL0w5rkQq54c,566
+emx_onnx_cgen/templates/gemm_op.c.j2,sha256=MK_hl_1EjjwL2QysAZ5oZqubayZDR7F2f9aDWGK0KJc,1492
+emx_onnx_cgen/templates/grid_sample_op.c.j2,sha256=xznm33Hne2SpQG7mCiGcnVuWe_NhmWcV4yOTdv04bQM,8938
+emx_onnx_cgen/templates/group_normalization_op.c.j2,sha256=vIxBXEAGmdrHmQ0p_-NjChf1eK5MODH1E5Zszf4x1ck,2297
+emx_onnx_cgen/templates/gru_op.c.j2,sha256=Mxv7-VrCsHZqN7DUIdPV6BDuBVHO90FkZb-MP9X0UFE,7247
+emx_onnx_cgen/templates/hamming_window_op.c.j2,sha256=pwwOd5fRuSEnlfTRkG5Bor5FiWvZagsJxaN87rJDpSs,562
+emx_onnx_cgen/templates/hardmax_op.c.j2,sha256=jfGo7aZRfs_rYh2RF0S-oUC6eW7wSlfSzsWNrNr4Sp8,1213
+emx_onnx_cgen/templates/identity_op.c.j2,sha256=cMIKmBjrY7mKNXWCPP96JzonYOLAyxvIhtur7u8QeoQ,410
+emx_onnx_cgen/templates/instance_normalization_op.c.j2,sha256=Ffj0S9UhfA0uR2bYS8xiJYmK3hKE7KrPfYRvCXdTvXM,1779
+emx_onnx_cgen/templates/layer_normalization_op.c.j2,sha256=M02O1ArelK1GP_E6sSd9s-SrO7hQerTey9xPbiWwZlE,3208
+emx_onnx_cgen/templates/logsoftmax_op.c.j2,sha256=xQZqD7wJELWV-QyC8J4Jn2LPspNPNxkOkzKiW1Nkwb4,1374
+emx_onnx_cgen/templates/lp_normalization_op.c.j2,sha256=d2RzG1aiCw7rEEik7BcC929xHBHlHA5TGMM9sUL8prQ,1134
+emx_onnx_cgen/templates/lp_pool_op.c.j2,sha256=hC91h4ynvh_4LGpkW6IE2H20wPpdkz65mKudasznhKA,1332
+emx_onnx_cgen/templates/lrn_op.c.j2,sha256=ANPey63qgozZYc4yXHkgo4eos08qxEIv0Bj30tZJTrw,972
+emx_onnx_cgen/templates/lstm_op.c.j2,sha256=B774zaw_MuNrWg7phs239e3YcD254WG2dOcg7b7-Mtk,8388
+emx_onnx_cgen/templates/matmul_op.c.j2,sha256=KbM-BE7opwN-qq7x83f9NvuLY0XSum9AhLI8aegDsQg,1028
+emx_onnx_cgen/templates/maxpool_op.c.j2,sha256=DOSbfeLqzJxVVfEq_mym7l1FtpWFJ4_SbGlMRU6dvsk,6733
+emx_onnx_cgen/templates/mean_variance_normalization_op.c.j2,sha256=fQofYSGnc4qYfo-bWHNliE8t4-pW4KdgSawMSyv2OiE,1408
+emx_onnx_cgen/templates/multi_input_op.c.j2,sha256=4cYUmzYX0r7Hby1iJys2Q9LiroQPIRqlbUiLIsMUxf4,663
+emx_onnx_cgen/templates/negative_log_likelihood_loss_op.c.j2,sha256=ECWSgma_vQWZNpV35v5q46VJkwPcm1JEAl7G7bDvqU4,1958
+emx_onnx_cgen/templates/nonmax_suppression_op.c.j2,sha256=-A9MHZQiOMIMyajbhl_sg3zsq8QjjA09nlFYdZF--8o,7574
+emx_onnx_cgen/templates/nonzero_op.c.j2,sha256=rmI_nV-s0fbUKzgNF92J5t1Ur0P4I9TsTAEnD6mjyeg,534
+emx_onnx_cgen/templates/one_hot_op.c.j2,sha256=frzEGK47nHL_rUJJtvOe2o1_5I1M_dcNkqKbr0AMQWY,1003
+emx_onnx_cgen/templates/optional_has_element_op.c.j2,sha256=Q1ydPD03IwBF4kOlQF1njzyzdUim_RWFkwrP3oPysio,145
+emx_onnx_cgen/templates/pad_op.c.j2,sha256=jmNJz9oDVmlKavuTsgLc2_Y1V-T6IZb8MgAcI7hEQiI,3426
+emx_onnx_cgen/templates/qlinear_matmul_op.c.j2,sha256=KJZH-aQEWIOIr_e8bQDMd5jZaUPdjBd6IdpgJRjNX0s,2898
+emx_onnx_cgen/templates/qlinear_mul_op.c.j2,sha256=KxTJ1W3TiZtg14Y2tY-RhalwvXF_JrAhoi0KvtTOfQs,1768
+emx_onnx_cgen/templates/quantize_linear_op.c.j2,sha256=IToa5IkdFxlFRKsJP8g8oM8iK-y5xcgRM6NCXq-8rpw,683
+emx_onnx_cgen/templates/range_op.c.j2,sha256=5fT0Ow-gr5KwGgKv7IuW6HTWD1SEHVyVxgPEI4jbpQg,343
+emx_onnx_cgen/templates/reduce_op.c.j2,sha256=tHGGkqz-1zJed9qlXFfMw3LUcWVssDWDw707oeStiNE,951
+emx_onnx_cgen/templates/reduce_op_dynamic.c.j2,sha256=hTb-OE2Q4e8us7iOESTcVxTdbvSUo3fimDxIuWrQ3iA,2741
+emx_onnx_cgen/templates/reshape_op.c.j2,sha256=ok-_aWeoFFtRUyCQmDpNH8hFEUr8yDL8xzUzyY2nm4I,717
+emx_onnx_cgen/templates/resize_op.c.j2,sha256=Z2DAGFV3o7y4UK6cStCDQKny9fYXMNyUaW30Z_wxrJw,10708
+emx_onnx_cgen/templates/rms_normalization_op.c.j2,sha256=jZap5WBXDIXOaouErQpsg2xOppefvNk16OBvxwrWIVY,1505
+emx_onnx_cgen/templates/rotary_embedding_op.c.j2,sha256=hx7wP9y-R4E8irzN4bVCNBUFKAJIIl_56fr2xB6ZFlI,2757
+emx_onnx_cgen/templates/scatter_nd_op.c.j2,sha256=zBRZhdl8t7V09jn9WYYoXzMfZQwipotxHZw9d9OUP1Y,1888
+emx_onnx_cgen/templates/shape_op.c.j2,sha256=1RAbUeK5qcNl_aClcccFfNJXHKgAUg3RLxcbZOQFvss,192
+emx_onnx_cgen/templates/size_op.c.j2,sha256=QFmXFo8GJ0J7X3Gk7AKkWJ1H8CMziMIex9yxh8vF390,137
+emx_onnx_cgen/templates/slice_op.c.j2,sha256=lzt-sLsu9_ceB3NeCJHbViGB3uGRPPdxjLuMoyJxpdA,416
+emx_onnx_cgen/templates/slice_op_dynamic.c.j2,sha256=qQ-c5JJA6haEtEgr4YAv_S0r65JOpza5xyxTRKSqSxk,2386
+emx_onnx_cgen/templates/softmax_cross_entropy_loss_op.c.j2,sha256=8svd2Cdxi2LI5B91rZB5hmpfeCorKy1t_yPS-wUXe5A,3907
+emx_onnx_cgen/templates/softmax_op.c.j2,sha256=kGi-Tfte70tpkdSmZEb3QqIGKCTYfmwExwF6UU0HVoo,1287
+emx_onnx_cgen/templates/space_to_depth_op.c.j2,sha256=URarlsQ2OQJWR1iFW9g4Id10ml9xGEI5Cfd2HLucZr8,1107
+emx_onnx_cgen/templates/split_op.c.j2,sha256=pTgMEim_WsTGRpkHmTrbp6MD0QUgbuWYNg0aXrLMPQE,1001
+emx_onnx_cgen/templates/tensor_scatter_op.c.j2,sha256=cWY6HRF45ndD5OJiudR_qJ5gA74e1Ni0KzEo8En8-lc,1708
+emx_onnx_cgen/templates/testbench.c.j2,sha256=a2gomUR6ZRwp1iEMf9cU26YCxx075RoKSSagrPbtSME,5098
+emx_onnx_cgen/templates/tfidf_vectorizer_op.c.j2,sha256=aOAdPywoj_YdQefmZM8m8I16LhE4cpvsB1dzeKz7tHs,6734
+emx_onnx_cgen/templates/tile_op.c.j2,sha256=4zGcRy6E5oDQosUVkVpv_gBae-tPoz3PuDBghwxVTUY,593
+emx_onnx_cgen/templates/topk_op.c.j2,sha256=BlIZrhasfYPflNhVSVL2u8KyJOvg2DAYkEs8rN9bAGs,2311
+emx_onnx_cgen/templates/transpose_op.c.j2,sha256=4fUCHukFNFNx0cd3HdjlsoE2_rf-VYM7Qy0n9F5NJzA,416
+emx_onnx_cgen/templates/trilu_op.c.j2,sha256=-3vIYEIg6hpkKF_cWZQUgWdwZDfBAhg82ObE0wr0mPQ,1302
+emx_onnx_cgen/templates/unary_op.c.j2,sha256=O2y4ZGtk0pcrUNRVCX3URm9l7BtYuM1PM4UPU5e-U3A,1615
+emx_onnx_cgen/templates/where_op.c.j2,sha256=p7D2exEzl15-EnHYyLyxeNsFjycUXm8AYU4zOJQChf8,357
+shared/__init__.py,sha256=bmP79AVZdY_1aNULJap9pm76Q41Rabrza6X-0A8lDzw,45
+shared/scalar_functions.py,sha256=CErro1Du2Ri3uqX6Dgd18DzNbxduckAvsmLJ6oHGx9A,91123
+shared/scalar_types.py,sha256=kEpsl5T-NVFxCcTzXqPJbtpvDiCgKHfz91dphLLZxZA,4912
+shared/ulp.py,sha256=DpeovCFijmP8_M7zyTZWsNyfOtJ1AjNSdxf5jGsdfJo,1856
+emx_onnx_cgen-0.4.1.dev0.dist-info/METADATA,sha256=Vc8-2kAZWsb85o-MH8--f6ofR9GBi75nqJ1FfmGAxk8,8185
+emx_onnx_cgen-0.4.1.dev0.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
+emx_onnx_cgen-0.4.1.dev0.dist-info/entry_points.txt,sha256=b7Rvmz_Bi9kWyn7QayQC_FEXiRpt4cS1RnluKh49yoo,57
+emx_onnx_cgen-0.4.1.dev0.dist-info/top_level.txt,sha256=g39fo-blEbgiVcC_GRqAnBzN234w3LXbcVdLUoItSLk,21
+emx_onnx_cgen-0.4.1.dev0.dist-info/RECORD,,

{emx_onnx_cgen-0.3.8.dist-info → emx_onnx_cgen-0.4.1.dev0.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (80.10.1)
+Generator: setuptools (80.10.2)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

emx_onnx_cgen/runtime/__init__.py

@@ -1 +0,0 @@
-"""Runtime helpers for evaluating ONNX graphs."""