From 15606309a05ccf7fadbaad5538cb7c32acb1e06b Mon Sep 17 00:00:00 2001 From: JidongZhang-THU <1119708529@qq.com> Date: Wed, 31 Jan 2024 21:10:15 +0800 Subject: llava : add MobileVLM support (#5132) * New Feature: 1. Sum_Rows: fix cuda kernel overflow fix block shape error when nrows too big 2. Im2Col: Support Batch in cuda Support f32 to f32 both in cpu && cuda 3. DepthWiseConv: Support by Im2Col && MulMat 4. Pool_2d: Supoort avg pooling in cuda 5. HardSigmoid: Imp in cuda 6. HardSwish: Imp in cuda * fix tabs instead of spaces * code clean * CUDA POOL2D * ADD POOL2D test case in test-backend-ops.cpp * code clean * fix pool2d_kernel nits * fix bug in pool2d kernel * fix avg pooling, count_include_pad nits * test-backend-ops : add more pool_2d tests * cuda : fix warnings and formatting * ggml : check types in release builds too in pool_2d * test-backend-ops : remove f16 pool_2d tests * cuda : more style fixes * Add assert in ggml_cuda_op_pool2d * pool2d float padding fallback * test-backend-ops : add dst_type to im2col --------- Co-authored-by: slaren --- examples/llava/MobileVLM-README.md | 58 ++++++++++++++++++++++++++++++++++++-- 1 file changed, 56 insertions(+), 2 deletions(-) (limited to 'examples') diff --git a/examples/llava/MobileVLM-README.md b/examples/llava/MobileVLM-README.md index c6258eba..9eba791d 100644 --- a/examples/llava/MobileVLM-README.md +++ b/examples/llava/MobileVLM-README.md @@ -111,17 +111,71 @@ llama_print_timings: eval time = 1279.03 ms / 18 runs ( 71.06 m llama_print_timings: total time = 34570.79 ms ``` +## Orin compile and run +### compile +```sh +make LLAMA_CUBLAS=1 CUDA_DOCKER_ARCH=sm_87 LLAMA_CUDA_F16=1 -j 32 +``` + +### run on Orin +### case 1 +**input** +```sh +./llava-cli \ + -m /data/local/tmp/ggml-model-q4_k.gguf \ + --mmproj /data/local/tmp/mmproj-model-f16.gguf \ + --image /data/local/tmp/demo.jpeg \ + -p "A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions. USER: \nWho is the author of this book? \nAnswer the question using a single word or phrase. ASSISTANT:" \ + --n-gpu-layers 999 +``` +**output** +```sh + +encode_image_with_clip: image encoded in 296.62 ms by CLIP ( 2.06 ms per image patch) + + Susan Wise Bauer + +llama_print_timings: load time = 1067.64 ms +llama_print_timings: sample time = 1.53 ms / 6 runs ( 0.25 ms per token, 3934.43 tokens per second) +llama_print_timings: prompt eval time = 306.84 ms / 246 tokens ( 1.25 ms per token, 801.72 tokens per second) +llama_print_timings: eval time = 91.50 ms / 6 runs ( 15.25 ms per token, 65.58 tokens per second) +llama_print_timings: total time = 1352.63 ms / 252 tokens +``` + +### case 2 +**input** +```sh +./llava-cli \ + -m /data/local/tmp/ggml-model-q4_k.gguf \ + --mmproj /data/local/tmp/mmproj-model-f16.gguf \ + -p "A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions. USER: \nWhat is in the image? ASSISTANT:" \ + --n-gpu-layers 999 + +``` +**output** +```sh +encode_image_with_clip: image encoded in 302.15 ms by CLIP ( 2.10 ms per image patch) + + The image features a cat lying in the grass. + +llama_print_timings: load time = 1057.07 ms +llama_print_timings: sample time = 3.27 ms / 11 runs ( 0.30 ms per token, 3360.83 tokens per second) +llama_print_timings: prompt eval time = 213.60 ms / 232 tokens ( 0.92 ms per token, 1086.14 tokens per second) +llama_print_timings: eval time = 166.65 ms / 11 runs ( 15.15 ms per token, 66.01 tokens per second) +llama_print_timings: total time = 1365.47 ms / 243 tokens +``` + ## Minor shortcomings The `n_patch` of output in `ldp` is 1/4 of the input. In order to implement quickly, we uniformly modified `clip_n_patches` function to a quarter. when counting the time consumption, the calculated time will be 4 times bigger than the real cost. ## TODO -- [ ] Support non-CPU backend for the new operators, such as `depthwise`, `hardswish`, `hardsigmoid` +- [x] Support non-CPU backend for the new operators, such as `depthwise`, `hardswish`, `hardsigmoid` - [ ] Optimize LDP projector performance - Optimize the structure definition to avoid unnecessary memory rearrangements, to reduce the use of `ggml_permute_cpy`; - Optimize operator implementation (ARM CPU/NVIDIA GPU): such as depthwise conv, hardswish, hardsigmoid, etc. -- [ ] run MobileVLM on `Jetson Orin` +- [x] run MobileVLM on `Jetson Orin` - [ ] Support more model variants, such as `MobileVLM-3B`. -- cgit v1.2.3