Commit
•
e605551
1
Parent(s):
b7ceb19
Update README.md
Browse files
README.md
CHANGED
|
@@ -126,6 +126,7 @@ python fms-extras/scripts/paged_speculative_inference.py \
|
|
| 126 |
--tokenizer=$MODEL_PATH \
|
| 127 |
--speculator_path=ibm-fms/llama-13b-accelerator \
|
| 128 |
--speculator_source=hf \
|
|
|
|
| 129 |
--compile \
|
| 130 |
--compile_mode=reduce-overhead
|
| 131 |
```
|
|
@@ -141,6 +142,7 @@ python fms-extras/scripts/paged_speculative_inference.py \
|
|
| 141 |
--tokenizer=$MODEL_PATH \
|
| 142 |
--speculator_path=ibm-fms/llama-13b-accelerator \
|
| 143 |
--speculator_source=hf \
|
|
|
|
| 144 |
--compile \
|
| 145 |
```
|
| 146 |
|
|
@@ -155,6 +157,7 @@ python fms-extras/scripts/paged_speculative_inference.py \
|
|
| 155 |
--tokenizer=$MODEL_PATH \
|
| 156 |
--speculator_path=ibm-fms/llama-13b-accelerator \
|
| 157 |
--speculator_source=hf \
|
|
|
|
| 158 |
--batch_input \
|
| 159 |
--compile \
|
| 160 |
```
|
|
|
|
| 126 |
--tokenizer=$MODEL_PATH \
|
| 127 |
--speculator_path=ibm-fms/llama-13b-accelerator \
|
| 128 |
--speculator_source=hf \
|
| 129 |
+
--speculator_variant=840m \
|
| 130 |
--compile \
|
| 131 |
--compile_mode=reduce-overhead
|
| 132 |
```
|
|
|
|
| 142 |
--tokenizer=$MODEL_PATH \
|
| 143 |
--speculator_path=ibm-fms/llama-13b-accelerator \
|
| 144 |
--speculator_source=hf \
|
| 145 |
+
--speculator_variant=840m \
|
| 146 |
--compile \
|
| 147 |
```
|
| 148 |
|
|
|
|
| 157 |
--tokenizer=$MODEL_PATH \
|
| 158 |
--speculator_path=ibm-fms/llama-13b-accelerator \
|
| 159 |
--speculator_source=hf \
|
| 160 |
+
--speculator_variant=840m \
|
| 161 |
--batch_input \
|
| 162 |
--compile \
|
| 163 |
```
|