Instructions to use Writer/Palmyra-Med-70B-32K with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Writer/Palmyra-Med-70B-32K with Transformers:

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

pipe = pipeline("text-generation", model="Writer/Palmyra-Med-70B-32K")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Writer/Palmyra-Med-70B-32K")
model = AutoModelForCausalLM.from_pretrained("Writer/Palmyra-Med-70B-32K")
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]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use Writer/Palmyra-Med-70B-32K with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Writer/Palmyra-Med-70B-32K"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Writer/Palmyra-Med-70B-32K",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Writer/Palmyra-Med-70B-32K

SGLang

How to use Writer/Palmyra-Med-70B-32K 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 "Writer/Palmyra-Med-70B-32K" \
    --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": "Writer/Palmyra-Med-70B-32K",
		"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 "Writer/Palmyra-Med-70B-32K" \
        --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": "Writer/Palmyra-Med-70B-32K",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Writer/Palmyra-Med-70B-32K with Docker Model Runner:
```
docker model run hf.co/Writer/Palmyra-Med-70B-32K
```

Base model?

by KevalRx - opened Aug 16, 2024

Discussion

KevalRx

Aug 16, 2024

•

edited Aug 16, 2024

What base model was used to train Palmyra-Med-70B? was it llama3-70B?

wassemgtk

Writer org Aug 16, 2024

Palmyra-004-70B

KevalRx

Aug 16, 2024

•

edited Aug 16, 2024

And the base model for Palmyra-004-70B is llama3-70B? I see llama listed in the model card, so curious which llama model did you use?

wassemgtk

Writer org Aug 17, 2024

It's a Llama-style model but not the Llama model, trained on a total of 4 trillion tokens.

cmcmaster

Aug 18, 2024

•

edited Aug 18, 2024

Oh, that makes a lot of sense. Physician here. This model has a lot of very substantial knowledge gaps. Why didn't you just finetune from a good, open weights model?

wassemgtk

Writer org Aug 18, 2024

Based on our evaluation, Palmyra-003 and 004 is outperforming Llama 2 and 3 across the board, so it made sense to use the stronger model as the base :)

https://dev.writer.com/home/models

wassemgtk changed discussion status to closed Sep 4, 2024

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment