Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,41 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: mit
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
pipeline_tag: text-generation
|
| 4 |
+
tags:
|
| 5 |
+
- ONNX
|
| 6 |
+
- DML
|
| 7 |
+
- ONNXRuntime
|
| 8 |
+
- phi3
|
| 9 |
+
- nlp
|
| 10 |
+
- conversational
|
| 11 |
+
- custom_code
|
| 12 |
+
inference: false
|
| 13 |
+
language:
|
| 14 |
+
- en
|
| 15 |
+
---
|
| 16 |
+
# EmbeddedLLM/Phi-3-mini-4k-instruct-onnx-cpu-int4-rtn-block-32-acc-level-4
|
| 17 |
+
|
| 18 |
+
## Performance Metrics
|
| 19 |
+
|
| 20 |
+
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 21 |
+
### CPU-INT4-RTN-BLOCK-32-ACC-LEVEL-4
|
| 22 |
+
We measured the performance of CPU-INT4-RTN-BLOCK-32-ACC-LEVEL-4 on AMD Ryzen 9 7940HS /w Radeon 78
|
| 23 |
+
|
| 24 |
+
| Prompt Length | Generation Length | Average Throughput (tps) |
|
| 25 |
+
|---------------------------|-------------------|-----------------------------|
|
| 26 |
+
| 128 | 128 | - |
|
| 27 |
+
| 128 | 256 | - |
|
| 28 |
+
| 128 | 512 | - |
|
| 29 |
+
| 128 | 1024 | - |
|
| 30 |
+
| 256 | 128 | - |
|
| 31 |
+
| 256 | 256 | - |
|
| 32 |
+
| 256 | 512 | - |
|
| 33 |
+
| 256 | 1024 | - |
|
| 34 |
+
| 512 | 128 | - |
|
| 35 |
+
| 512 | 256 | - |
|
| 36 |
+
| 512 | 512 | - |
|
| 37 |
+
| 512 | 1024 | - |
|
| 38 |
+
| 1024 | 128 | - |
|
| 39 |
+
| 1024 | 256 | - |
|
| 40 |
+
| 1024 | 512 | - |
|
| 41 |
+
| 1024 | 1024 | - |
|