 myhloli
|
2f1369a877
feat: add Mac environment checks and support for Apple Silicon in backend selection
|
2 semanas atrás |
|
myhloli
|
6a6c650062
fix: correct spelling of set_default_gpu_memory_utilization and set_default_batch_size functions
|
3 semanas atrás |
|
myhloli
|
0d0ebfd7bc
fix: improve GPU memory utilization handling and ensure OMP_NUM_THREADS is set only if not defined
|
3 semanas atrás |
|
myhloli
|
078aaaf150
fix: remove unnecessary parameters from kwargs in vlm_analyze.py initialization
|
3 semanas atrás |
|
myhloli
|
c3a55fffab
fix: add utility functions for GPU memory utilization and batch size configuration
|
3 semanas atrás |
|
myhloli
|
0758de6d24
Update vllm version and increase default GPU memory utilization
|
4 semanas atrás |
|
myhloli
|
ea4271ab37
Set OMP_NUM_THREADS environment variable to 1 for vllm backend initialization
|
1 mês atrás |
|
myhloli
|
4b480cfcf7
Add max_concurrency parameter to improve backend processing
|
1 mês atrás |
|
myhloli
|
f1659eb7a7
Refactor logits processor handling in server.py and vlm_analyze.py for improved clarity and consistency
|
1 mês atrás |
|
myhloli
|
7c481796f8
Refactor custom logits processors to include vllm version checks and improve logging
|
1 mês atrás |
|
myhloli
|
f5e0e67545
Add custom logits processors functionality with compute capability check
|
1 mês atrás |
|
myhloli
|
a4cac624df
Add compute capability check for custom logits processors in server.py and vlm_analyze.py
|
1 mês atrás |
|
myhloli
|
9fc3d6c742
Remove direct import of MinerULogitsProcessor and add it conditionally in vllm backend
|
1 mês atrás |
|
myhloli
|
72f1f5f935
Update mineru-vl-utils version and add logits processors support
|
1 mês atrás |
|
myhloli
|
3ca520a3fe
feat: implement dynamic batch size calculation based on GPU memory in vlm_analyze.py
|
2 meses atrás |
|
myhloli
|
11ebb47891
fix: remove redundant model_path checks for vllm backends in vlm_analyze.py
|
2 meses atrás |
|
myhloli
|
7a71cfe288
feat: add support for vllm-async-engine backend in vlm_analyze.py
|
2 meses atrás |
|
myhloli
|
c4eae4e0ef
fix: add timing log for model predictor retrieval in vlm_analyze.py
|
2 meses atrás |
|
myhloli
|
411f3b7855
fix: comment out debug logging in vllm_analyze.py
|
2 meses atrás |
|
myhloli
|
60e257e5f1
fix: set default values for gpu_memory_utilization and model in vllm_analyze.py
|
2 meses atrás |
|
myhloli
|
20e1dfe984
fix: enhance model initialization for transformers and vllm-engine backends in vlm_analyze.py
|
2 meses atrás |
|
myhloli
|
55eaad224d
feat: add support for vlm 2.5
|
2 meses atrás |
|
myhloli
|
29754f7c17
fix: update image loading to support BASE64 format for improved compatibility
|
3 meses atrás |
|
myhloli
|
ff11e602fc
feat: enhance image processing by introducing ImageType enum and updating related functions
|
3 meses atrás |
|
myhloli
|
9a3a314916
refactor: add support for additional keyword arguments in client and analysis functions
|
4 meses atrás |
|
myhloli
|
38637a6a35
feat: allow additional keyword arguments in model loading function
|
4 meses atrás |
|
myhloli
|
e7649e432c
refactor: modularize PDF processing and enhance async support for VLM backend
|
4 meses atrás |
|
myhloli
|
ec3adde809
fix: remove unused import of ModelPath in vlm_analyze.py
|
5 meses atrás |
|
myhloli
|
72c946a9ec
fix: include model_path in key for backend model retrieval in vlm_analyze.py
|
5 meses atrás |
|
myhloli
|
20dcbd2164
fix: include server_url in model key for backend model retrieval in vlm_analyze.py
|
5 meses atrás |
