branch: master
Commits on master
- e78df48 update inputs for transfers in hsagraph (#3560) 2 years ago
- 086291e hotfix: add test for JIT reset 2 years ago
- dccefab remove mixtral weight to clang first (#3792) 2 years ago
- bf3e1c4 support pickling tensors and others (#3787) 2 years ago
- 5ac1fa9 apply the same fix_bf16 in llama and coder (#3789) 2 years ago
- 639bd5d move bf16 cast hack to Tensor.llvm_bf16_cast (#3788) 2 years ago
- 311cf2b Revert "threefry_2x32 (#2601)" (#3784) 2 years ago
- db3de54 threefry_2x32 (#2601) 2 years ago
- 53adcb3 remove hip backend (#3783) 2 years ago
- 2a14d1b Revert "add outbufs info to CompiledASTRunner (#3781)" (#3782) 2 years ago
- 722dd42 add outbufs info to CompiledASTRunner (#3781) 2 years ago
- 9255332 use llvm as bridge to fix_bf16 loading (#3774) 2 years ago
- 987a055 increase jit batch size progressivly (#3771) 2 years ago
- 77febb4 llama 7B on 6 gpus benchmark (#3773) 2 years ago
- 07324b5 [experimenting] use contiguous instead of realize in optim (#3770) 2 years ago
- e3e89c2 multioutput uoping infra (#3706) 2 years ago
- e1c5aa9 estimated resnet training time for BENCHMARK (#3769) 2 years ago
- 0870dd5 hotfix: switch resnet training from HIP -> HSA in CI 2 years ago
- d807087 bfs scheduler, infra for multioutput (#3763) 2 years ago
- 91e181e make alignment readable (#3766) 2 years ago
- 8ea5395 bfloat16 Tensor.rand (#3764) 2 years ago
- a2d3cf6 move is_dtype_supported to test.helpers (#3762) 2 years ago
- 8af87e2 unrealized, assign is replace (#3761) 2 years ago
- 922f831 Run test_real_world in METAL test (#3760) 2 years ago
- 4bd5535 update mlperf resnet default hparams (#3758) 2 years ago
- aad9332 remove that extra assign line, is it fixed? (#3757) 2 years ago
- ba79a3c some hsa lines saving + fixes (#3752) 2 years ago
- ca19eb3 where fold try 2 (#3748) 2 years ago
- 6b8c66e fix broken loops in llvm (#3751) 2 years ago
- d3a6319 bf16 tests in test_dtype.py (#3749) 2 years ago