v0.1.2

What's Changed

• [RWKV7] fix RWKV7Attention.__init__ by @exhyy in #238
• fix(triton): remove num_warps=8 in bwd_prepare_wy_repr_kernel to avoid MMA layout assertion on non-Ampere GPUs. by @kugwzk in #240
• [Fix]: reshape o before o_proj in linear_attn layer. by @Luther-Sparks in #243
• [CI] Seperate tests to compile , normal and varlen by @zhiyuan1i in #247
• [ABC] Add use_rope parameter to ABCAttention and ABCConfig & Fix compiler bugs in kernels by @yzhangcs in #248
• [CI] trigger GPU workflow only on pull_request events by @zhiyuan1i in #249
• Create test_linearatten.py by @kangyiyang in #250
• [CI] Fix all erros and enable testing for PR by @zhiyuan1i in #251
• [CI] add H100 GPU by @zhiyuan1i in #254
• [Gated DeltaNet] fix gdn kernel bugs on h100 when vdim=64 by @kugwzk in #256
• [Test] Enhance support for NVIDIA Hopper GPU by @zhiyuan1i in #257
• [FAQ] Update triton-nightly links by @yzhangcs in #259
• [Attn] Add triton impls for MHA/GQA by @yzhangcs in #260
• [Attn] Use larger block size for hopper devices by @yzhangcs in #261
• [Attn] Enable test for attn by @zhiyuan1i in #262
• [CI] fix a syntax error in triton-nightly by @zhiyuan1i in #263
• Bump fla to v0.1.2 by @yzhangcs in #264

New Contributors

• @exhyy made their first contribution in #238
• @kugwzk made their first contribution in #240
• @Luther-Sparks made their first contribution in #243
• @yzhangcs made their first contribution in #248
• @kangyiyang made their first contribution in #250

Full Changelog: v0.1.1...v0.1.2