Instructions to use kalo-team/llama3-4x8b-pythonT2_step_final with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use kalo-team/llama3-4x8b-pythonT2_step_final with Transformers:

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

pipe = pipeline("text-generation", model="kalo-team/llama3-4x8b-pythonT2_step_final")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("kalo-team/llama3-4x8b-pythonT2_step_final")
model = AutoModelForCausalLM.from_pretrained("kalo-team/llama3-4x8b-pythonT2_step_final")
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

vLLM

How to use kalo-team/llama3-4x8b-pythonT2_step_final with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "kalo-team/llama3-4x8b-pythonT2_step_final"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "kalo-team/llama3-4x8b-pythonT2_step_final",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/kalo-team/llama3-4x8b-pythonT2_step_final

SGLang

How to use kalo-team/llama3-4x8b-pythonT2_step_final 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 "kalo-team/llama3-4x8b-pythonT2_step_final" \
    --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": "kalo-team/llama3-4x8b-pythonT2_step_final",
		"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 "kalo-team/llama3-4x8b-pythonT2_step_final" \
        --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": "kalo-team/llama3-4x8b-pythonT2_step_final",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use kalo-team/llama3-4x8b-pythonT2_step_final with Docker Model Runner:
```
docker model run hf.co/kalo-team/llama3-4x8b-pythonT2_step_final
```

kalomaze commited on May 22, 2024

Commit

e7b159b

verified ·

1 Parent(s): e7766e1

Update README.md

Browse files

Files changed (1) hide show

README.md +7 -0

README.md CHANGED Viewed

@@ -1,3 +1,10 @@
 # 70b Distillation Experiment
 This is not the full-fledged run that I plan to do for a large scale distillation of Llama3 70b.
 Instead, it's a preliminary test train of the custom distillation trainer, where we target KL divergence from the larger Llama3 70b teacher model onto 4x8b (the student).

+---
+license: llama3
+language:
+- en
+tags:
+- code
+---
 # 70b Distillation Experiment
 This is not the full-fledged run that I plan to do for a large scale distillation of Llama3 70b.
 Instead, it's a preliminary test train of the custom distillation trainer, where we target KL divergence from the larger Llama3 70b teacher model onto 4x8b (the student).