Hugging Face's logo

ibm-ai-platform
/
codellama-13b-accelerator

Transformers
Safetensors
mlp_speculator
Model card Files
xet
Community
JRosenkranz commited on
Commit
b956eb0
·
verified ·
1 Parent(s): 8e0f096

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +3 -3
README.md CHANGED
@@ -138,7 +138,7 @@ pip install transformers==4.35.0 sentencepiece numpy
138
  ```bash
139
  MODEL_PATH=/path/to/llama/hf/CodeLlama-13b-Instruct-hf
140
  python fms-extras/scripts/paged_speculative_inference.py \
141
- --variant=13b_code \
142
  --model_path=$MODEL_PATH \
143
  --model_source=hf \
144
  --tokenizer=$MODEL_PATH \
@@ -156,7 +156,7 @@ python fms-extras/scripts/paged_speculative_inference.py \
156
  ```bash
157
  MODEL_PATH=/path/to/llama/hf/CodeLlama-13b-Instruct-hf
158
  python fms-extras/scripts/paged_speculative_inference.py \
159
- --variant=13b_code \
160
  --model_path=$MODEL_PATH \
161
  --model_source=hf \
162
  --tokenizer=$MODEL_PATH \
@@ -173,7 +173,7 @@ python fms-extras/scripts/paged_speculative_inference.py \
173
  ```bash
174
  MODEL_PATH=/path/to/llama/hf/CodeLlama-13b-Instruct-hf
175
  python fms-extras/scripts/paged_speculative_inference.py \
176
- --variant=13b_code \
177
  --model_path=$MODEL_PATH \
178
  --model_source=hf \
179
  --tokenizer=$MODEL_PATH \
 
138
  ```bash
139
  MODEL_PATH=/path/to/llama/hf/CodeLlama-13b-Instruct-hf
140
  python fms-extras/scripts/paged_speculative_inference.py \
141
+ --variant=13b.code \
142
  --model_path=$MODEL_PATH \
143
  --model_source=hf \
144
  --tokenizer=$MODEL_PATH \
 
156
  ```bash
157
  MODEL_PATH=/path/to/llama/hf/CodeLlama-13b-Instruct-hf
158
  python fms-extras/scripts/paged_speculative_inference.py \
159
+ --variant=13b.code \
160
  --model_path=$MODEL_PATH \
161
  --model_source=hf \
162
  --tokenizer=$MODEL_PATH \
 
173
  ```bash
174
  MODEL_PATH=/path/to/llama/hf/CodeLlama-13b-Instruct-hf
175
  python fms-extras/scripts/paged_speculative_inference.py \
176
+ --variant=13b.code \
177
  --model_path=$MODEL_PATH \
178
  --model_source=hf \
179
  --tokenizer=$MODEL_PATH \