caiwanxianhust / FasterLLaMALinks
使用 CUDA C++ 实现的 llama 模型推理框架
☆62Updated 11 months ago
Alternatives and similar repositories for FasterLLaMA
Users that are interested in FasterLLaMA are comparing it to the libraries listed below
Sorting:
- A light llama-like llm inference framework based on the triton kernel.☆158Updated last month
- ☆141Updated last year
- 使用 cutlass 实现 flash-attention 精简版,具 有教学意义☆50Updated last year
- llm theoretical performance analysis tools and support params, flops, memory and latency analysis.☆108Updated 3 months ago
- 使用 cutlass 仓库在 ada 架构上实现 fp8 的 flash attention☆76Updated last year
- ☆150Updated 9 months ago
- 分层解耦的深度学习推理引擎☆76Updated 8 months ago
- ☆107Updated 5 months ago
- Performance of the C++ interface of flash attention and flash attention v2 in large language model (LLM) inference scenarios.☆41Updated 7 months ago
- ☆33Updated 8 months ago
- Decoding Attention is specially optimized for MHA, MQA, GQA and MLA using CUDA core for the decoding stage of LLM inference.☆45Updated 4 months ago
- ☆139Updated last year
- ☆59Updated 11 months ago
- ☆56Updated 3 months ago
- ☆14Updated 7 months ago
- ☆36Updated 5 months ago
- 🤖FFPA: Extend FlashAttention-2 with Split-D, ~O(1) SRAM complexity for large headdim, 1.8x~3x↑🎉 vs SDPA EA.☆223Updated 2 months ago
- GEMV implementation with CUTLASS☆13Updated 2 months ago
- ☆44Updated last year
- Implement Flash Attention using Cute.☆96Updated 10 months ago
- ⚡️Write HGEMM from scratch using Tensor Cores with WMMA, MMA and CuTe API, Achieve Peak⚡️ Performance.☆121Updated 5 months ago
- A practical way of learning Swizzle☆29Updated 8 months ago
- ☆18Updated last year
- Examples of CUDA implementations by Cutlass CuTe☆241Updated 3 months ago
- Optimize softmax in triton in many cases☆21Updated last year
- 注释的nano_vllm仓库,并且完成了MiniCPM4的适配以及注册新模型的功能☆81Updated 2 months ago
- Standalone Flash Attention v2 kernel without libtorch dependency☆112Updated last year
- Tutorials for writing high-performance GPU operators in AI frameworks.☆132Updated 2 years ago
- Several optimization methods of half-precision general matrix vector multiplication (HGEMV) using CUDA core.☆67Updated last year
- A standalone GEMM kernel for fp16 activation and quantized weight, extracted from FasterTransformer☆94Updated last month