Instructions to use waldie/Yi-34B-GiftedConvo-merged-4bpw-h6-exl2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use waldie/Yi-34B-GiftedConvo-merged-4bpw-h6-exl2 with Transformers:

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

pipe = pipeline("text-generation", model="waldie/Yi-34B-GiftedConvo-merged-4bpw-h6-exl2")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("waldie/Yi-34B-GiftedConvo-merged-4bpw-h6-exl2")
model = AutoModelForCausalLM.from_pretrained("waldie/Yi-34B-GiftedConvo-merged-4bpw-h6-exl2", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use waldie/Yi-34B-GiftedConvo-merged-4bpw-h6-exl2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "waldie/Yi-34B-GiftedConvo-merged-4bpw-h6-exl2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "waldie/Yi-34B-GiftedConvo-merged-4bpw-h6-exl2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/waldie/Yi-34B-GiftedConvo-merged-4bpw-h6-exl2

SGLang

How to use waldie/Yi-34B-GiftedConvo-merged-4bpw-h6-exl2 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 "waldie/Yi-34B-GiftedConvo-merged-4bpw-h6-exl2" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "waldie/Yi-34B-GiftedConvo-merged-4bpw-h6-exl2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "waldie/Yi-34B-GiftedConvo-merged-4bpw-h6-exl2" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "waldie/Yi-34B-GiftedConvo-merged-4bpw-h6-exl2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use waldie/Yi-34B-GiftedConvo-merged-4bpw-h6-exl2 with Docker Model Runner:
```
docker model run hf.co/waldie/Yi-34B-GiftedConvo-merged-4bpw-h6-exl2
```

Trained on over 20k instruct generated all by gpt-4 or humans

Dataset features: 1000 long evolved conversations based off LIMA Subsection of correct PRM800k data Subsection of CamelAI's Physics and Chemistry data

The model is trained with Qlora as well as Axolotl.

Downloads last month: 10

waldie
/

Yi-34B-GiftedConvo-merged-4bpw-h6-exl2

Dataset used to train waldie/Yi-34B-GiftedConvo-merged-4bpw-h6-exl2