myhloli
|
64586d03ea
fix: import MinerULogitsProcessor conditionally for vllm-engine and vllm-async-engine backends
|
5 days ago |
myhloli
|
cf1fbd2923
fix: enhance device and backend configuration handling in lmdeploy and vlm modules
|
1 week ago |
myhloli
|
0a4c9e307f
fix: enhance http-client backend parameters in vlm_analyze.py for improved configuration options
|
2 weeks ago |
myhloli
|
d2c93b770f
fix: refactor backend handling in vlm_analyze.py for improved model loading and error handling
|
2 weeks ago |
myhloli
|
3293299f34
fix: update README to clarify Windows LMDeploy backend performance and compatibility
|
2 weeks ago |
myhloli
|
39e7aa52a2
fix: improve device type and backend handling in lmdeploy configuration
|
2 weeks ago |
myhloli
|
a2136c22a5
fix: add backend argument handling and logging for lmdeploy backend configuration
|
2 weeks ago |
myhloli
|
3fcca35c73
fix: add Linux environment detection and set lmdeploy backend based on device type
|
2 weeks ago |
myhloli
|
5c743dc169
fix: update device handling and backend configuration in analysis scripts
|
2 weeks ago |
myhloli
|
85558061ff
feat: add lmdeploy backend support and refactor related components
|
2 weeks ago |
jinminxi104
|
9ec03c0353
add lmdeploy-backend
|
3 weeks ago |
myhloli
|
4afa045545
refactor: update import statement to use check_sys_env and adjust logging level for image loading
|
3 weeks ago |
myhloli
|
14c334d2b0
feat: add macOS version check for mlx-engine backend support
|
1 month ago |
myhloli
|
2f1369a877
feat: add Mac environment checks and support for Apple Silicon in backend selection
|
1 month ago |
myhloli
|
6a6c650062
fix: correct spelling of set_default_gpu_memory_utilization and set_default_batch_size functions
|
1 month ago |
myhloli
|
0d0ebfd7bc
fix: improve GPU memory utilization handling and ensure OMP_NUM_THREADS is set only if not defined
|
1 month ago |
myhloli
|
078aaaf150
fix: remove unnecessary parameters from kwargs in vlm_analyze.py initialization
|
1 month ago |
myhloli
|
c3a55fffab
fix: add utility functions for GPU memory utilization and batch size configuration
|
1 month ago |
myhloli
|
0758de6d24
Update vllm version and increase default GPU memory utilization
|
1 month ago |
myhloli
|
ea4271ab37
Set OMP_NUM_THREADS environment variable to 1 for vllm backend initialization
|
1 month ago |
myhloli
|
4b480cfcf7
Add max_concurrency parameter to improve backend processing
|
1 month ago |
myhloli
|
f1659eb7a7
Refactor logits processor handling in server.py and vlm_analyze.py for improved clarity and consistency
|
2 months ago |
myhloli
|
7c481796f8
Refactor custom logits processors to include vllm version checks and improve logging
|
2 months ago |
myhloli
|
f5e0e67545
Add custom logits processors functionality with compute capability check
|
2 months ago |
myhloli
|
a4cac624df
Add compute capability check for custom logits processors in server.py and vlm_analyze.py
|
2 months ago |
myhloli
|
9fc3d6c742
Remove direct import of MinerULogitsProcessor and add it conditionally in vllm backend
|
2 months ago |
myhloli
|
72f1f5f935
Update mineru-vl-utils version and add logits processors support
|
2 months ago |
myhloli
|
3ca520a3fe
feat: implement dynamic batch size calculation based on GPU memory in vlm_analyze.py
|
2 months ago |
myhloli
|
11ebb47891
fix: remove redundant model_path checks for vllm backends in vlm_analyze.py
|
2 months ago |
myhloli
|
7a71cfe288
feat: add support for vllm-async-engine backend in vlm_analyze.py
|
2 months ago |