Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
* update dlrmv2 BKC (#1476) * use merge-emb-cat for int8 since acc issue is fixed in IPEX (#1477) * forcing numpy to use a specific version (#1474) * Restrict vit training to single socket (#1484) * update dlrm/bert distribute training BKC (#1486) * change batch size for int8 resnet50 and ssd-resnet34 (#1487) * Added command to remove existing logs in output_dir (#1475) * Added command to remove existing logs in output_dir * Fix condition to checkout OneAPI tools repository (#1490) * Hz/dlrm ddp (#1496) * fix dlrm ddp * fix time computation --------- Co-authored-by: Weizhuo Zhang <[email protected]> * fix dlrm-v1 int8 thp (#1497) * also use merged-emb-cat in dlrm-v2 int8 thp (#1498) * updated tpp files for 2.12.1 release (#1479) * updated tpp files * added yolo5 * P0 models list (#1500) * P0 models list * replace master w/ tag * correct framework name --------- Co-authored-by: Jitendra Patil <[email protected]> * another update to TPPs (#1503) * Fixing SSD-Resnet34 training quickstart script to run right number of instances (#1493) * Container GHA Pipeline Reformat (#1462) * swap runner to mlops runner * change ipex base image (#1440) * rename tests yaml (#1450) * New Test Runner (#1461) * add execute perms to quickstart and add 140 tests to pytorch resnet with new runner * add tests per new format * add flex140 support * Update Test Runner (#1467) * Flex 140 tests for P0 (#1469) * add previous m3 commits (#1478) * GHA tests for flex 140 (#1499) * add previous m3 commits (#1478) * Added command to remove existing logs in output_dir (#1475) * address PR review (#1501) * remove makefile * Remove caas reference (#1502) * Add previous m3 commits in baremetal readme (#1480) --------- Co-authored-by: Srikanth Ramakrishna <[email protected]> Co-authored-by: mahathis <[email protected]> * refine dlrm ddp dataloader (#1504) Co-authored-by: Weizhuo Zhang <[email protected]> * workaround oneccl bad termination issue for RN50 distributed training (#1508) * Fix Test Pipeline (#1514) * fix test pipeline * Update container-pipeline-tester.yml * Bump mlflow in /datasets/cloud_data_connector/samples/interoperability (#1492) Bumps [mlflow](https://github.com/mlflow/mlflow) from 2.5.0 to 2.6.0. * Bump mlflow in /datasets/cloud_data_connector/samples/azure (#1491) Bumps [mlflow](https://github.com/mlflow/mlflow) from 2.5.0 to 2.6.0. * Zufang/readme update for itex (#1485) * add link to int8 PB for onednn graph * refine readme for onednn graph option * MaskRCNN GPU training (#1513) * maskmrcnn model demo zero-bkc * update readme * added license header * added HW requirement * update docs * support bf32 for SD finetune (#1521) * fix dlrm-v1 ddp hang (#1512) * fix dlrm-v1 ddp hang * comment out two more barrier * modify multi-node scripts for resnet50, maskrcnn and stable diffusion (#1537) * modify multi-node scripts for RNNT and ssd-resnet34 (#1539) * Modify Test Runner Run Dir from GHA (#1541) * Adjust test runner path to be MLOps root * move to test-runner dir * merge dir and parent_dir * remove parent dir * use full paths * get test name for artifact upload * Stop workload tests and security scans on open PR event (#1543) * PVC P0 RN50 PYT Inference (#1494) * updated to latest BKC to add multi-card multi-tile support * PVC P0 PYT BERT Large (#1495) * Add support for multi-card multi-tile * PVC P0 PYT DLRM training (#1518) * add dlrm training pvc support * Fix LKG pipeline for the new ipex and itex conda installers (#1525) * update LKG pipeline for the new ipex and itex conda installers * remove version subdir from bom file * TF RN50V1_5 P0 Max Inference (#1529) Co-authored-by: Mahathi Vatsal <[email protected]> * modify SD inference scripts (#1553) * Updates + New Transfer Learning Notebooks from TLT team (#1522) * Update for all transfer learning notebooks --------- Co-authored-by: Harsha Ramayanam <[email protected]> * Rename the repo to Intel AI Reference Models for rebranding (#1473) * Rename the repo for rebranding, remove k8s and tools directories * Update README.md --------- Co-authored-by: Clayne Robison <[email protected]> * modify SD finetune scripts (#1555) * fix CrossEntropyLoss target for dummy inputs (#1556) * PVC P0 PYT DLRM inference (#1517) * add dlrm pvc inference support * Modified baremetal README(bert and rn50) for max series (#1520) * fix itex (#1527) * fix itex * fix key error * PVC BERT-Large P0 TF (#1534) * adapt new BKC * Modified Bert large TF baremetal readme for Mx series (#1557) --------- Co-authored-by: Mahathi Vatsal <[email protected]> * Add optimizations for BF16 transformer inference (#1489) * Change the order of operations from dense->concat->split_heads to dense->split_heads->concat (attention_layer.py) * Change the order of operations when calculating encoder-decoder k, v caches to avoid using matmul ops between large matrices(transformer.py) *Reduce number of occurrences of split_heads with encoder-decoder k, v caches by performing split_heads in transformer.py * Update README.md (#1516) Change BFloat16 model description to point to new frozen graph. * make llama training max-step flexible (#1563) * Fix drops logic to be avoided if any workload test fails (#1564) * Create selective PR validations tags-based (#1549) * Create selective PR validations tags-based * Add edit_pull_request as trigger for the PR validations --------- Co-authored-by: Wafaa Taie <[email protected]> * PYT CPU Automation (#1544) * make initial changes * add tests for new base container * add more new tests * remove env var * add more tests * more test added * add dlrm inference build and test * add more tests * add more tests * add another model test * add final tests * add devcatalog (#1566) * Change success condition from previous jobs before doing drop (#1568) * Added readme for DLRM pytorch MAX series (#1570) * update docs for AI Tools (#1567) * ViT Train : Enable multi instance training for Tensorflow Vision Transformer model (#1569) * Revert "Restrict vit training to single socket (#1484)" This reverts commit 8d4eb7a. * Add multi instance support * Update README.md * Remove useless files and add license title for DLRM v2 (#1574) * Gda/step url (#1560) * Added step url to result table * Remove continue on error from workload tests * Run performance checks even when a workload test failed * add driver setup doc (#1571) Co-authored-by: Jitendra Patil <[email protected]> * LLM models using ipex.optimize_transformers for bf16/int8 (#1562) * init pr * revise v1, local test passed * TF ResNet50v1.5 Fix (#1575) * tf cpu r50 inf fixed * actions.json added * newline added for actions * venv instllation added * pip install fixed * venv pip install fixed * test workflow reverted back --------- Co-authored-by: Jitendra Patil <[email protected]> * added int8 support for graphsage (#1536) * MEMREC DLRM Inference (#1547) * Added omp_num_threads and cores_per_instance as env variable (#1573) * esnet50v1.5 and bert large inference models * TF Stable Diffusion: Download model files in start.sh and log latency & throughput (#1581) * adding model download to start.sh to avoid failure during multi-instance execution * downloading the clip tokenizer in start.sh also * script changes to report both latency and throughput * Doc fixes (#1582) * add torchrec dlrm to the models table * fix paths to run quickstart scripts * Add RN50 INT8 Calibration file (#1545) * Jupyter notebook for AI Reference models (#1583) * Added jupyter notebook for AI Reference models * Added README for AI_Reference jupyter notebook * Supports Resnet50 v1.5 and mobilenet v1 inference workloads * Upgrade Pillow version to 10.0.1 to fix high severity CVEs (#1584) * remove workflows * correct_release_tag (#1587) * correct_release_tag * revert a change * Unexpose old TF models (#1593) * tf cpu distilbert inf (#1612) * 3D Unet MLPerf Inference Workload (#1595) * 3D Unet MLPerf added * docker compose added * batch size corrected * numactl added * ubuntu Dockerfile updated * output dir changed in tests * yes flag added in Dockerfile * default OS added * BERT Large Inference CPU Workload Added (#1594) * BERT Large Inf added * TCMALLOC added * ubuntu Dockerfile updated * TCMalloc location updated * test file updated * yes flag added in Dockerfile * TF CPU Bert Large Training Workload (#1596) * bert large pretraining added * extra OS removed from r50 inf service * ubuntu Dockerfile updated * ssh helper script added * yum non tineractive update * output dir fixed * TF CPU DIEN Inference Workload (#1598) * TF CPU MobileNet V1 Inference Workload (#1600) * TF CPU ResNet v1.5 Training Workload (#1604) * TF CPU SSD ResNet-34 Inference Workload (#1606) * TF CPU SSD ResNet-34 Training Workload (#1607) * TF CPU Transformer MLPerf Inference Workload (#1597) * TF CPU Transformer MLPerf Training (#1608) * TF CPU DistilBERT fixed (#1629) * TF CPU SSD MobileNet Inference Workload (#1605) * Fixed typo in readme for framework (#1631) * Checkpoints added for TF CPU Workloads (#1637) * TF CPU Dev Catalog READMEs Updated. (#1652) * EMR PYT RN50 Infer (#1624) --------- Co-authored-by: Jitendra Patil <[email protected]> * EMR PYT RN50 Train (#1625) * build rn50 train centos * comment conda lines * comment conda lines * remove fp16 test,add devcat and intel-openmp * add changes to ubuntu * remove commented lines * add cpu to tag * EMR PYT ResNext Infer (#1619) * add initial commits for emr resnext * add dockerfiles * build resnext * remove extra precisions * add devcat,openmp and more tests * add cpu to tag * EMR PYT MaskRCNN Infer (#1621) * add maskrcnn inference * add cmake and bf32 tests * add openmpi,more tests and devcat * add cpu to tag * rearrange pip installs * EMR PYT MaskRCNN Train (#1622) * build maskrcnn training * add cmake * correct image names * add ld_predload,devcat and tests changes * EMR PYT SSD-ResNet34 Train (#1618) * add compose changes * ssd-resnet34 build * correct tests file * add more tests and devcat * EMR PYT SSD-ResNet34 Infer (#1617) * build ssd-resnet34 images * add bf32 tests and update DEVCATALOG.md * rename devcatalogs (#1671) * EMR PYT BERT Large Infer (#1623) * add bert-large build * correct paths * add devcatalog * add more tests * Rename EMR_DEVCATALOG.md to DEVCATALOG.md * Update DEVCATALOG.md * PYT EMR BERT-Large Train (#1639) * build bert-large training * add pretrained model env * remove idsid * add more tests and devcatalog * correct env and rename * Delete EMR_DEVCATALOG.md * Update DEVCATALOG.md * EMR PYT Distilbert Infer (#1620) * build distilbert images * validate distilbert * add more tests and devcatalog * remove MZ reference * modify env params * uncomment and remove idsid * clarify core per instance * clarify hf_datasets * remove void env var * Rename EMR_DEVCATALOG.md to DEVCATALOG.md * Update DEVCATALOG.md * EMR PYT RNNT Inference (#1616) * add dockerfiles for rnnt * fix pytorch binding error * copy diff file to inference * add librosa * add more tests and devcatalog * correct reatime cmd * Rename EMR_DEVCATALOG.md to DEVCATALOG.md * Update DEVCATALOG.md * EMR PYT RNNT Train (#1615) * Rename EMR_DEVCATALOG.md to DEVCATALOG.md * Update DEVCATALOG.md * EMR PYT DLRM Infer (#1626) * Update DEVCATALOG.md * EMR PYT DLRM Train (#1628) * build dlrm training * add num_batch * add tcmalloc * add tcmalloc * add devcatalog * re-locate the file * Rename EMR_DEVCATALOG.md to DEVCATALOG.md * Update DEVCATALOG.md * make batch flexible (#1635) * Change dataset for Transfer Learning LLM Notebook (#1576) * update llm notebook with code alpaca * push updates * fixed broken link * Refactor Transfer Learning Notebook folder to match TLT structure + small diff (#1670) * refactor to match TL structure + small diff * fixed table structure * add landing page doc (#1653) * add landing page doc * simplify and add r3.1 * add precisions * add tf landing page * add precisions --------- Co-authored-by: Jitendra Patil <[email protected]> * r3.1 fixes (#1679) * TF CPU ResNet 50 v1.5 Inference Model Checkpoints fixed (#1663) * spr removed from workdir * R50 Inf fixed * fixed rn50 error * fixed in docker compose yaml --------- Co-authored-by: Sharvil Shah <[email protected]> * make minor corrections in devcatalog README (#1680) * Remove old TF models (#1673) * remove ResNet50, FasterCNN, RFCN, NCF, Wide and deep Large dataset training, waveNet, Inception v4, mlperf GNMT models * remove relevant unit tests and update coverage precentage * refine changes based on feedback (#1684) * fix typo (#1686) * fixing docker iamages names * fixing TF centos docker images link * unset KMP AFFINITY for accuracy scripts (#1689) * Bump mlflow in /datasets/cloud_data_connector/samples/azure (#1698) Bumps [mlflow](https://github.com/mlflow/mlflow) from 2.6.0 to 2.8.1. - [Release notes](https://github.com/mlflow/mlflow/releases) - [Changelog](https://github.com/mlflow/mlflow/blob/master/CHANGELOG.md) - [Commits](mlflow/mlflow@v2.6.0...v2.8.1) --- updated-dependencies: - dependency-name: mlflow dependency-type: direct:production * Bump mlflow in /datasets/cloud_data_connector/samples/interoperability (#1697) Bumps [mlflow](https://github.com/mlflow/mlflow) from 2.6.0 to 2.8.1. - [Release notes](https://github.com/mlflow/mlflow/releases) - [Changelog](https://github.com/mlflow/mlflow/blob/master/CHANGELOG.md) - [Commits](mlflow/mlflow@v2.6.0...v2.8.1) --- updated-dependencies: - dependency-name: mlflow dependency-type: direct:production * add optional args to devcatalog pages (#1750) * add optional env for tf * add optional args * add v2 to table (#1762) * remove space * Corrected updating OMP NUM THREADS (#1759) * validate omp_num_threads and cores_per_instance * revert changes * validate omp num threads and cores per instance (#1789) * add omp_num_threads and cores_per_instance (#1809) * wsl2 documentation (#1815) * add wsl2 stable diffusion documentation * add wsl2 stable diffusion documentation * add wsl2 base doc * make minor tabular changes * make minor tabular changes * add ssh instructions * re-word example * update torch version outputs * add entry in main readme --------- Co-authored-by: Jitendra Patil <[email protected]> * fix dlrm normal training to support SNC mode (#1699) Co-authored-by: Weizhuo Zhang <[email protected]> * update torch-ccl branch (#1716) * Pytorch ResNext32x16d baremetal EMR tests (#1640) * Pytorch RN50 baremetal EMR inference and training tests (#1641) * PyTorch SSD-Resnet34 EMR baremetal training and inference tests (#1647) * PyTorch distilBERT baremetal tests (#1650) * PyTorch BERT_LARGE_SQUAD inf baremetal tests (#1651) * PyTorch BERT_LARGE Training baremetal tests (#1654) * Pytorch MaskRCNN baremetal EMR tests (#1646) * PyTorch DLRM baremetal tests (#1649) * PyTorch RNN-T EMR baremetal tests (#1648) * remove TF yolo v5, add cpu stable diffusion * [Zero-BKC][ITEX][GPU]add itex stable diffusion,EfficientNet,wide and deep inference for ats-m (#1642) Co-authored-by: XumingGai <[email protected]> * remove pb files from the github repo (#1730) * modify rn50 training script (#1732) * Refactor new zero bkcs scripts for TF ResNet50 inf and Mask-RCNN to models_v2 (#1695) * move new zero bkcs for TF resnet50 inf and maskrcnn to models_v2 * GHA tests for dGPU zero copy BKC format workloads (#1744) --------- Co-authored-by: Mahathi Vatsal <[email protected]> * modify Stable Diffusion finetune script (#1746) * add utilities to parse result for pytorch (#1729) * Wliao2/add rn50 (#1693) * add dgpu resnet 50 * update intel tf version to be the latest (#1748) * Enable Inductor path for Bert_large inference and training (#1733) * Init Bert-large files from inductor path * cherry pick Enable int8-mixed-bf16 for 5 transformer models (#1720) * modify README * Enable Inductor path for Distilbert-base inference (#1734) * Init Bert-large files from inductor path * cherry pick Enable int8-mixed-bf16 for 5 transformer models (#1720) * modify README * Init Distilbert base models * Init DLRM Script (#1739) * Enable Inductor path for RN50 inference and training (#1718) * Enable Inductor path for RN50 inference and training * add bf32 * add README for Torch inductor --------- Co-authored-by: leslie-fang-intel <[email protected]> * Move and add license headers P1 ATS-M ITEX models (#1754) * move and add license headers to stable diffusion model * move and add license headers to efficientnet * update maskrcnn inference * move and update license headers for wide and deep model * Remove MLFlow dependency. Updates on functional tests (#1756) * add v2 to table (#1762) * Weizhuoz/fix bert accuracy (#1761) * fix bert-large accuracy read issue * fix bert_large accuracy issue * inductor int8 could not use model.eval() * remove ssdmobilenet and yolov4 (#1757) * update maskrcnn, bert-large training (#1666) * update maskrcnn training * update maskrcnn and add bert-large * move maskrcnn training to model_v2, update license * code review changes for bert large training --------- Co-authored-by: Wafaa Taie <[email protected]> * update mlflow version to use the latest (#1768) * [Zero-BKC][ITEX][GPU] Add resnet50 and 3dd-unet training (#1664) * add 3d-unet gpu training * add gpu resnet50 training * update license headers and move scripts to models_v2 * code review changes for 3d-unet training, update license, move to models_v2 * changes in readme for code review --------- Co-authored-by: Wafaa Taie <[email protected]> * enable distributed training for DLRMv2 and some fix for inductor path (#1770) * enable distributed training for DLRMv2 and some fix for inductor path * add missing files * remove license headers from .txt files in data-connector (#1774) * fix data loader (#1780) * Fix for TL Notebooks GHA (#1773) * fixed file paths * changed venv creation * trying venv * trying no venv * reverted venv3 * testing apt get update * added apt-get install * venv3 --> venv * uncommented apt * without virtualenv * added pip install virtualenv * downgraded PyYaml to be conpatible with tf models official * 2.12.0 --> 2.12.1 * removed package versions * added back tf official version * flipped * addressed review comments * Fix inductor path int8 bf16 realtime issue (#1767) * Update inference_performance.sh (#1784) * fix acc (#1798) * add warm up iter for inference throughput (#1799) * Predownload weights for SDv2.1 (#1797) * predownload weights for SDv2.1 * update hash * fix DLRM V1 syntax error (#1760) * fix DLRM V1 syntax error * fix dlrm inductor numpy.bool_ error (#1448) * fix DLRM V1 train_ld issue * Update dlrm_s_pytorch.py --------- Co-authored-by: Chunyuan WU <[email protected]> Co-authored-by: zhuhaozhe <[email protected]> * Fix bert large inductor int8 accuray failure in last batch (#1803) * fix bert large inductor int8 accuracy issue * Format fixes --------- Co-authored-by: jianan-gu <[email protected]> * add bert (#1701) * add bert IPEX Co-authored-by: Wafaa Taie <[email protected]> Co-authored-by: Mahathi Vatsal <[email protected]> * Wliao2/add dlrm kaggle (#1711) * add dlrm kaggle * fix license issue * Update README.md * remove unused files * update * add licence header for modified files --------- Co-authored-by: Mahathi Vatsal <[email protected]> * update bert-large for ARC (#1816) * update scripts and Readme for ARC * Update README.md * Update quickstart scripts with env variables for Stable Diffusion and ResNet50v1.5 (#1791) * adding env vars for SD and RN50 * updating accuracy quickstart and other minor changes * using default values only if the env var are not set from the cmd line * fix for coverage tests * Added GHA for ITEX wide deep large (#1820) * Added GHA for ITEX wide deep large * Added stable diffusion inference ITEX (#1819) * Added stable diffusion inference * Changed file permissions * Added GHA for EfficientNet ITEX (#1821) * Added GHA for EfficientNet ITEX * Update run_test.sh * Update setup.sh * Update README.md * Added GHA for ITEX bert large Training (#1822) * ssdmobilenet int8 accuracy fix (#1811) * ssdmobilenet int8 accuracy fix * added change in quickstart accuracy script * modified public bucket link * modified new args in unit test * fixed unit test * add BKC for DLRM-V2 convergence test (#1824) * wsl2 documentation (#1815) * add wsl2 stable diffusion documentation * add wsl2 stable diffusion documentation * add wsl2 base doc * make minor tabular changes * make minor tabular changes * add ssh instructions * re-word example * update torch version outputs * add entry in main readme --------- Co-authored-by: Jitendra Patil <[email protected]> * bert-large inductor uses int8_bf16 mix (#1792) * Bert-large inductor use int8-bf16 mix * ipex uses int8_bf16 mix in if * merge develop * inductor uses int8-bf16 mix * Distilbert int8 optimization (#1830) * optimize distilbert int8 * re-calibrate distilbert * fix for inductor (#1834) * TF- DistilBERT - Update quickstart scripts with env vars (#1818) * Add Env variables to quickstart scripts * Update # of cores for throughput script * add distilbert (#1702) * add distilbert * Corrected refactored path for distilbert * Added intel license header * Update README.md --------- Co-authored-by: Mahathi Vatsal <[email protected]> * Update dlrm_s_pytorch.py (#1843) * Wliao2/add stable diffusion (#1705) * add stable_diffusion * update some typo * fix license issue * update stable diffusion * update for acc * verify the result * Refactored to new folder * Update README.md * add support for ARC --------- Co-authored-by: Mahathi Vatsal <[email protected]> * Fix Bert Large Int8 Latency Issue (#1859) Co-authored-by: jianan-gu <[email protected]> * [DistilBert] modify for masked_fill default value (#1868) * Nhatle/bert large training x3 vs x1 (#1776) * set num-inter-threads=2 * bert-large squad: Binding process to cores on 1 socket * Enable multi-instance training for bert-large squad * Fix incase users only run 1 instance * Fix benchmark_command * Molly/ddp bkc update (#1873) * make num_iter flexbile * bugfix for bert-large ddp * bkc for rn50 ddp training update * bkc for rn50 ddp training update * bkc for dlrm_v1 ddp training update --------- Co-authored-by: WeizhuoZhang-intel <[email protected]> * Corrected IPEX installer version (#1878) * Changed IPEX installer version * Update dlrm_s_pytorch.py (#1879) * Update AI Bundle version in tests setup files for CI/CD pipeline (#1881) * doc: document models_v2 contribution guideline (#1855) Signed-off-by: Dmitry Rogozhkin <[email protected]> * Molly/inductor fp16 (#1875) * make num_iter flexbile * bugfix for bert-large ddp * bkc for rn50 ddp training update * bkc for rn50 ddp training update * bkc for dlrm_v1 ddp training update * rn50 fp16 torch.compile enabled * fp16 autocast fix * fix RN50 for fp16 torch.compile (#1849) * enable stable-diffusion fp16 inductor path * vit, bert-large fp16 enable * merge to latest transformers patch * Update enable_ipex_for_transformers.diff * Update enable_ipex_for_transformers.diff --------- Co-authored-by: Cao E <[email protected]> Co-authored-by: WeizhuoZhang-intel <[email protected]> * Fix in case mpi_num_processes_per_socket=1 (#1885) * Fix in case mpi_num_processes_per_socket=1 * small fix * Update dlrm_s_pytorch.py (#1890) * Modified dataset path (#1894) * added GHA for ITEX bert large * Stable Diffusion PYT Flex and Max (#1853) * validate sd pyt * add max tests and dockerfile * Bump scipy in /models_v2/pytorch/stable_diffusion/inference/gpu (#1847) Bumps [scipy](https://github.com/scipy/scipy) from 1.9.1 to 1.11.1. - [Release notes](https://github.com/scipy/scipy/releases) - [Commits](scipy/scipy@v1.9.1...v1.11.1) --- updated-dependencies: - dependency-name: scipy dependency-type: direct:production * Bump gitpython in /models_v2/pytorch/distilbert/inference/gpu (#1846) Bumps [gitpython](https://github.com/gitpython-developers/GitPython) from 3.1.30 to 3.1.41. - [Release notes](https://github.com/gitpython-developers/GitPython/releases) - [Changelog](https://github.com/gitpython-developers/GitPython/blob/main/CHANGES) - [Commits](gitpython-developers/GitPython@3.1.30...3.1.41) --- updated-dependencies: - dependency-name: gitpython dependency-type: direct:production * Bump transformers in /models_v2/pytorch/distilbert/inference/gpu (#1840) Bumps [transformers](https://github.com/huggingface/transformers) from 4.25.1 to 4.36.0. - [Release notes](https://github.com/huggingface/transformers/releases) - [Commits](huggingface/transformers@v4.25.1...v4.36.0) --- updated-dependencies: - dependency-name: transformers dependency-type: direct:production * Bump transformers in /models_v2/pytorch/bert_large/inference/gpu (#1810) Bumps [transformers](https://github.com/huggingface/transformers) from 4.11.0 to 4.36.0. - [Release notes](https://github.com/huggingface/transformers/releases) - [Commits](huggingface/transformers@v4.11.0...v4.36.0) --- updated-dependencies: - dependency-name: transformers dependency-type: direct:production * Bump transformers (#1786) Bumps [transformers](https://github.com/huggingface/transformers) from 4.30.0 to 4.36.0. - [Release notes](https://github.com/huggingface/transformers/releases) - [Commits](huggingface/transformers@v4.30.0...v4.36.0) --- updated-dependencies: - dependency-name: transformers dependency-type: direct:production * Bump transformers (#1785) Bumps [transformers](https://github.com/huggingface/transformers) from 4.30.0 to 4.36.0. - [Release notes](https://github.com/huggingface/transformers/releases) - [Commits](huggingface/transformers@v4.30.0...v4.36.0) --- updated-dependencies: - dependency-name: transformers dependency-type: direct:production * Added GHA tests for stable diffusion (#1887) * validate sd pyt * update for ARC (#1781) * update for ARC * update log * refactor to models_v2 * update path due to refactor * sync with 2.1rc3 * Wliao2/add dlrm (#1704) * add dlrm v2 * fix license issue * Update dlrm_dataloader.py * Update dist_models.py * Update dlrm_dataloader.py * Update dist_models.py * refactored to a new folder * update Readme --------- Co-authored-by: Mahathi Vatsal <[email protected]> * Max 3D-Unet container support (#1832) * add masrkcnn container support * add 3d-unet container support * Added GHA for 3d Unet Training ITEX (#1823) * Added Resnet50v1.5 and maskrcnn train GHA test (#1751) * Refactored resnet50v1_5 for Zero copy BKC format (#1897) * Added Necessary Metadata and Bug Fixes for Transfer Learning Notebooks (#1825) * fixed file paths * changed venv creation * added pip install virtualenv * downgraded PyYaml to be conpatible with tf models official * 2.12.0 --> 2.12.1 * added back tf official version * addressed review comments * fixed version for fsspec, removed llm test * changed accelerate version * fix for tf-models-official * added needed metadata * made tests significantly less expensive * fixed zip extract to tar extract * fixed sms download * fixed typo for csv path name * addressed review comments * simplified if statement * Added oneapi path (#1902) * validate sd pyt * Update README.md with ipex version (#1903) * Update README.md with ipex version * Max MaskRCNN container support (#1831) * add masrkcnn container support * Max RN50 container validation (#1829) * validate container for zero-bkc for rn50 max container * Max BERT-Large container support (#1833) * add bert-large container support * Flex Wide and deep container (#1851) * validate zero-copy bkc for itex stable diffusion * validate zero-copy bkc for flex container (#1772) * validate zero-copy bkc * EfficientNet Container for flex (#1771) * validate zero-copy bkc efficientnet * TF MaskRCNN container for Flex GPU (#1755) * adapt zero-copy bkc and validate maskrcnn * validate bert-large inference PYT PVC (#1841) * validate bert-large inference * validate bert-large container PVC pytorch (#1838) * validate bert-large container * RN50 PYT Max container (#1904) * validate refactor of zero-bkc training * Latest updates to TF RN50 for Flex series (#1813) * adapt zero-copy bkc for image build * Update README.md for IPEX versions (#1907) * Update README.md * not cast and ramdomrized crossnet bias for inductor and make warmup iters as an arg (#1906) * resolve merge conflicts (#1911) * Wliao2/add ssdmbv1 (#1817) * add ssd-mobilenetv1 Co-authored-by: Mahathi Vatsal <[email protected]> * add IPEX Max 3dunet (#1706) * add 3dunet for IPEX for 3DUnet for Max series * Updated baremetal readme for distil bert IPEX (#1908) * Updated baremetal readme distil bert for IPEX * DistilBERT inference container PYT Flex and Max (#1854) * add functional support * docker: fix broken docker-compose.yml (#1913) Fixes: 473d3b3 ("DistilBERT inference container PYT Flex and Max (#1854)") Signed-off-by: Dmitry Rogozhkin <[email protected]> * docker/flex: fix build and run for tf maskrcnn (#1896) Signed-off-by: Dmitry Rogozhkin <[email protected]> * Flex PYT DLRM-v1 inference (#1895) * build dlrmv1 container * Updated baremetal readme for DLRM v1 (#1909) * Updated baremetal readme for DLRM v1 * Update README.md (#1914) * remove extra test (#1916) * release docs for containers (#1915) * Update release container table * Updated main README table (#1919) * Updated main README table * clean up workflows * restore git submodules --------- Signed-off-by: Dmitry Rogozhkin <[email protected]> Co-authored-by: jianan-gu <[email protected]> Co-authored-by: zhuhaozhe <[email protected]> Co-authored-by: Om Thakkar <[email protected]> Co-authored-by: sachinmuradi <[email protected]> Co-authored-by: Cao E <[email protected]> Co-authored-by: mahathis <[email protected]> Co-authored-by: lerealno <[email protected]> Co-authored-by: DiweiSun <[email protected]> Co-authored-by: zengxian <[email protected]> Co-authored-by: Weizhuo Zhang <[email protected]> Co-authored-by: Jitendra Patil <[email protected]> Co-authored-by: Srikanth Ramakrishna <[email protected]> Co-authored-by: Mahmoud Abuzaina <[email protected]> Co-authored-by: Tyler Titsworth <[email protected]> Co-authored-by: jiayisunx <[email protected]> Co-authored-by: zofia <[email protected]> Co-authored-by: Mahathi Vatsal <[email protected]> Co-authored-by: okhleif-IL <[email protected]> Co-authored-by: Harsha Ramayanam <[email protected]> Co-authored-by: Clayne Robison <[email protected]> Co-authored-by: jianyizh <[email protected]> Co-authored-by: nhatle <[email protected]> Co-authored-by: gera-aldama <[email protected]> Co-authored-by: Real Novo, Luis <[email protected]> Co-authored-by: Sharvil Shah <[email protected]> Co-authored-by: Ashiq Imran <[email protected]> Co-authored-by: Gopi Krishna Jha <[email protected]> Co-authored-by: leslie-fang-intel <[email protected]> Co-authored-by: Sharvil Shah <[email protected]> Co-authored-by: Nick Camarena <[email protected]> Co-authored-by: xiangdong <[email protected]> Co-authored-by: wenjun liu <[email protected]> Co-authored-by: XumingGai <[email protected]> Co-authored-by: wincent8 <[email protected]> Co-authored-by: Jesus Herrera Ledon <[email protected]> Co-authored-by: XumingGai <[email protected]> Co-authored-by: Chunyuan WU <[email protected]> Co-authored-by: Syed Shahbaaz Ahmed <[email protected]> Co-authored-by: Xuan Liao <[email protected]> Co-authored-by: Dmitry Rogozhkin <[email protected]>
- Loading branch information
41 people
authored
Jan 27, 2024
1 parent
a36b1c2
commit 36d8b40