85.4 kB

Ctrl+K

2 contributors

Add the s256 and s512 graphs for float16 hardware: gliner25_decide_s256_npu_wfp16.tflite (Qualcomm NPU) and gliner25_decide_s512_fp16hw_wfp16.tflite (GPU FP16 with FP32 accumulation), with their card rows, contract entries, NOTICE sentence, conversion scripts and hfmodels.json

600fe62 verified 21 days ago

README.md

6.28 kB
Add the s256 and s512 graphs for float16 hardware: gliner25_decide_s256_npu_wfp16.tflite (Qualcomm NPU) and gliner25_decide_s512_fp16hw_wfp16.tflite (GPU FP16 with FP32 accumulation), with their card rows, contract entries, NOTICE sentence, conversion scripts and hfmodels.json 21 days ago
build_tables.py

6.35 kB
GLiNER2.5-Decide LiteRT: fp16-weight graphs s128/s256/s512 (+fp32 references), fp16/fp32 embedding tables, host contract, Python runtime, conversion scripts, Android sample, card 30 days ago
common.py

1.98 kB
GLiNER2.5-Decide LiteRT: fp16-weight graphs s128/s256/s512 (+fp32 references), fp16/fp32 embedding tables, host contract, Python runtime, conversion scripts, Android sample, card 30 days ago
converter_layout.py

647 Bytes
GLiNER2.5-Decide LiteRT: fp16-weight graphs s128/s256/s512 (+fp32 references), fp16/fp32 embedding tables, host contract, Python runtime, conversion scripts, Android sample, card 30 days ago
decide_graph.py

3.84 kB
GLiNER2.5-Decide LiteRT: fp16-weight graphs s128/s256/s512 (+fp32 references), fp16/fp32 embedding tables, host contract, Python runtime, conversion scripts, Android sample, card 30 days ago
export_npu.py

5.38 kB
Add the s256 and s512 graphs for float16 hardware: gliner25_decide_s256_npu_wfp16.tflite (Qualcomm NPU) and gliner25_decide_s512_fp16hw_wfp16.tflite (GPU FP16 with FP32 accumulation), with their card rows, contract entries, NOTICE sentence, conversion scripts and hfmodels.json 21 days ago
export_s128.py

4.89 kB
GLiNER2.5-Decide LiteRT: fp16-weight graphs s128/s256/s512 (+fp32 references), fp16/fp32 embedding tables, host contract, Python runtime, conversion scripts, Android sample, card 30 days ago
export_shape.py

3.88 kB
GLiNER2.5-Decide LiteRT: fp16-weight graphs s128/s256/s512 (+fp32 references), fp16/fp32 embedding tables, host contract, Python runtime, conversion scripts, Android sample, card 30 days ago
litert_cpu.py

4.26 kB
GLiNER2.5-Decide LiteRT: fp16-weight graphs s128/s256/s512 (+fp32 references), fp16/fp32 embedding tables, host contract, Python runtime, conversion scripts, Android sample, card 30 days ago
ln_scales_gliner25_decide_s128_npu_split_fc1_fp32.json

896 Bytes
Add gliner25_decide_s128_npu_wfp16.tflite, the s128 graph for float16 hardware (Qualcomm NPU), with its card section, contract entry, NOTICE line and conversion scripts 22 days ago
ln_scales_gliner25_decide_s256_npu_split_fc1_fp32.json

1.06 kB
Add the s256 and s512 graphs for float16 hardware: gliner25_decide_s256_npu_wfp16.tflite (Qualcomm NPU) and gliner25_decide_s512_fp16hw_wfp16.tflite (GPU FP16 with FP32 accumulation), with their card rows, contract entries, NOTICE sentence, conversion scripts and hfmodels.json 21 days ago
ln_scales_gliner25_decide_s512_fp16hw_split_fc1_fp32.json

1.06 kB
Add the s256 and s512 graphs for float16 hardware: gliner25_decide_s256_npu_wfp16.tflite (Qualcomm NPU) and gliner25_decide_s512_fp16hw_wfp16.tflite (GPU FP16 with FP32 accumulation), with their card rows, contract entries, NOTICE sentence, conversion scripts and hfmodels.json 21 days ago
make_hero_app.py

6.66 kB
GLiNER2.5-Decide LiteRT: fp16-weight graphs s128/s256/s512 (+fp32 references), fp16/fp32 embedding tables, host contract, Python runtime, conversion scripts, Android sample, card 30 days ago
npu_graph.py

7.19 kB
Add gliner25_decide_s128_npu_wfp16.tflite, the s128 graph for float16 hardware (Qualcomm NPU), with its card section, contract entry, NOTICE line and conversion scripts 22 days ago
profile_ranges.py

6.28 kB
Add gliner25_decide_s128_npu_wfp16.tflite, the s128 graph for float16 hardware (Qualcomm NPU), with its card section, contract entry, NOTICE line and conversion scripts 22 days ago
quantize_npu.py

8.21 kB
Add the s256 and s512 graphs for float16 hardware: gliner25_decide_s256_npu_wfp16.tflite (Qualcomm NPU) and gliner25_decide_s512_fp16hw_wfp16.tflite (GPU FP16 with FP32 accumulation), with their card rows, contract entries, NOTICE sentence, conversion scripts and hfmodels.json 21 days ago
quantize_wfp16.py

7.78 kB
GLiNER2.5-Decide LiteRT: fp16-weight graphs s128/s256/s512 (+fp32 references), fp16/fp32 embedding tables, host contract, Python runtime, conversion scripts, Android sample, card 30 days ago
shaped_layer.py

3.71 kB
GLiNER2.5-Decide LiteRT: fp16-weight graphs s128/s256/s512 (+fp32 references), fp16/fp32 embedding tables, host contract, Python runtime, conversion scripts, Android sample, card 30 days ago
wfp16_audit.py

5.05 kB
GLiNER2.5-Decide LiteRT: fp16-weight graphs s128/s256/s512 (+fp32 references), fp16/fp32 embedding tables, host contract, Python runtime, conversion scripts, Android sample, card 30 days ago