Instructions to use modulora-repoducibility/llama-30b-4bit-quantized with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use modulora-repoducibility/llama-30b-4bit-quantized with Transformers:

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

pipe = pipeline("text-generation", model="modulora-repoducibility/llama-30b-4bit-quantized")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("modulora-repoducibility/llama-30b-4bit-quantized")
model = AutoModelForCausalLM.from_pretrained("modulora-repoducibility/llama-30b-4bit-quantized")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use modulora-repoducibility/llama-30b-4bit-quantized with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "modulora-repoducibility/llama-30b-4bit-quantized"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "modulora-repoducibility/llama-30b-4bit-quantized",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/modulora-repoducibility/llama-30b-4bit-quantized

SGLang

How to use modulora-repoducibility/llama-30b-4bit-quantized 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 "modulora-repoducibility/llama-30b-4bit-quantized" \
    --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": "modulora-repoducibility/llama-30b-4bit-quantized",
		"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 "modulora-repoducibility/llama-30b-4bit-quantized" \
        --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": "modulora-repoducibility/llama-30b-4bit-quantized",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use modulora-repoducibility/llama-30b-4bit-quantized with Docker Model Runner:
```
docker model run hf.co/modulora-repoducibility/llama-30b-4bit-quantized
```

osbm commited on May 16, 2024

Commit

51c7cbc

verified ·

1 Parent(s): b432d37

Upload 3 files

Browse files

Files changed (3) hide show

config.json +29 -0
quant_config.json +11 -0
quantized_weights.pt +3 -0

config.json ADDED Viewed

	@@ -0,0 +1,29 @@

+{
+  "_name_or_path": "huggyllama/llama-30b",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "hidden_act": "silu",
+  "hidden_size": 6656,
+  "initializer_range": 0.02,
+  "intermediate_size": 17920,
+  "max_position_embeddings": 2048,
+  "max_sequence_length": 2048,
+  "model_type": "llama",
+  "num_attention_heads": 52,
+  "num_hidden_layers": 60,
+  "num_key_value_heads": 52,
+  "pad_token_id": 0,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 10000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.34.0",
+  "use_cache": true,
+  "vocab_size": 32000
+}

quant_config.json ADDED Viewed

	@@ -0,0 +1,11 @@

+{
+  "dataset": "c4",
+  "bits": 4,
+  "nsamples": 128,
+  "groupsize": 64,
+  "act_order": true,
+  "percdamp": 0.01,
+  "seed": 0,
+  "nearest": false,
+  "save": "./llama-30b-4bit-quantized"
+}

quantized_weights.pt ADDED Viewed

	@@ -0,0 +1,3 @@

+version https://git-lfs.github.com/spec/v1
+oid sha256:36f45e96508717659c25307b9deafe20f0b6230c2e0327a7aadadd6450e2f7d9
+size 18173183861