Instructions to use ssmits/Zamba2-1.2B-instruct-Dutch with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ssmits/Zamba2-1.2B-instruct-Dutch with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="ssmits/Zamba2-1.2B-instruct-Dutch")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("ssmits/Zamba2-1.2B-instruct-Dutch")
model = AutoModelForCausalLM.from_pretrained("ssmits/Zamba2-1.2B-instruct-Dutch")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use ssmits/Zamba2-1.2B-instruct-Dutch with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ssmits/Zamba2-1.2B-instruct-Dutch"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ssmits/Zamba2-1.2B-instruct-Dutch",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ssmits/Zamba2-1.2B-instruct-Dutch

SGLang

How to use ssmits/Zamba2-1.2B-instruct-Dutch with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "ssmits/Zamba2-1.2B-instruct-Dutch" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ssmits/Zamba2-1.2B-instruct-Dutch",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "ssmits/Zamba2-1.2B-instruct-Dutch" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ssmits/Zamba2-1.2B-instruct-Dutch",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use ssmits/Zamba2-1.2B-instruct-Dutch with Docker Model Runner:
```
docker model run hf.co/ssmits/Zamba2-1.2B-instruct-Dutch
```

ssmits commited on Nov 3, 2024

Commit

90f46ea

verified ·

1 Parent(s): 17db2ea

Upload 2 files

Browse files

Files changed (2) hide show

finetune.py +3 -1
optimize_lr.py +4 -1

finetune.py CHANGED Viewed

@@ -1,4 +1,5 @@
 import os
 CONTEXT_WINDOW = 1024 #has to fit in 4090
 HF_TOKEN = os.getenv("HF_TOKEN")
@@ -10,6 +11,7 @@ from transformers import (
 import torch
 from datasets import load_dataset
 from huggingface_hub import login
 # setup tokenizer
 tokenizer = AutoTokenizer.from_pretrained("Zyphra/Zamba2-1.2B-instruct", token=HF_TOKEN)
@@ -73,7 +75,7 @@ training_args = TrainingArguments(
     save_steps=500,
     save_total_limit=2,
     logging_steps=100,
-    learning_rate=2e-5,
     weight_decay=0.01,
     fp16=False,
     bf16=True,

 import os
+import json
 CONTEXT_WINDOW = 1024 #has to fit in 4090
 HF_TOKEN = os.getenv("HF_TOKEN")
 import torch
 from datasets import load_dataset
 from huggingface_hub import login
+from optimize_lr import best_lr
 # setup tokenizer
 tokenizer = AutoTokenizer.from_pretrained("Zyphra/Zamba2-1.2B-instruct", token=HF_TOKEN)
     save_steps=500,
     save_total_limit=2,
     logging_steps=100,
+    learning_rate=best_lr,
     weight_decay=0.01,
     fp16=False,
     bf16=True,

optimize_lr.py CHANGED Viewed

@@ -398,4 +398,7 @@ plot_gpr_results(study, final_optimization)
 # Save all results
 with open("lr_optimization_results.json", "w") as f:
-    json.dump(results, f, indent=4)

 # Save all results
 with open("lr_optimization_results.json", "w") as f:
+    json.dump(results, f, indent=4)
+# Store best learning rate as a variable for finetune.py to use
+best_lr = study.best_params["learning_rate"]