Instructions to use nm-testing/tinyllama-one-shot-static-quant-test-compressed with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nm-testing/tinyllama-one-shot-static-quant-test-compressed with Transformers:

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

pipe = pipeline("text-generation", model="nm-testing/tinyllama-one-shot-static-quant-test-compressed")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("nm-testing/tinyllama-one-shot-static-quant-test-compressed")
model = AutoModelForCausalLM.from_pretrained("nm-testing/tinyllama-one-shot-static-quant-test-compressed")
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 nm-testing/tinyllama-one-shot-static-quant-test-compressed with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nm-testing/tinyllama-one-shot-static-quant-test-compressed"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nm-testing/tinyllama-one-shot-static-quant-test-compressed",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/nm-testing/tinyllama-one-shot-static-quant-test-compressed

SGLang

How to use nm-testing/tinyllama-one-shot-static-quant-test-compressed 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 "nm-testing/tinyllama-one-shot-static-quant-test-compressed" \
    --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": "nm-testing/tinyllama-one-shot-static-quant-test-compressed",
		"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 "nm-testing/tinyllama-one-shot-static-quant-test-compressed" \
        --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": "nm-testing/tinyllama-one-shot-static-quant-test-compressed",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use nm-testing/tinyllama-one-shot-static-quant-test-compressed with Docker Model Runner:
```
docker model run hf.co/nm-testing/tinyllama-one-shot-static-quant-test-compressed
```

sadkins65 commited on May 22, 2024

Commit

7a68152

verified ·

1 Parent(s): a8e06b9

Upload folder using huggingface_hub

Browse files

Files changed (3) hide show

config.json +3 -4
model.safetensors +2 -2
recipe.yaml +2 -2

config.json CHANGED Viewed

@@ -39,9 +39,8 @@
         }
       },
       "format": "int-quantized",
-      "global_compression_ratio": 1.454005842904474,
       "ignore": [
-        "model.layers.0.mlp.down_proj",
         "lm_head"
       ],
       "quant_method": "sparseml",
@@ -49,7 +48,7 @@
     },
     "sparsity_config": {
       "format": "dense",
-      "global_sparsity": 7.713156824200198,
       "registry_requires_subclass": false,
       "sparsity_structure": "0:0"
     }
@@ -69,7 +68,7 @@
   "rope_scaling": null,
   "rope_theta": 10000.0,
   "tie_word_embeddings": false,
-  "torch_dtype": "bfloat16",
   "transformers_version": "4.39.0",
   "use_cache": true,
   "vocab_size": 32000

         }
       },
       "format": "int-quantized",
+      "global_compression_ratio": 1.4583332397611055,
       "ignore": [
         "lm_head"
       ],
       "quant_method": "sparseml",
     },
     "sparsity_config": {
       "format": "dense",
+      "global_sparsity": 7.8259900429979625,
       "registry_requires_subclass": false,
       "sparsity_structure": "0:0"
     }
   "rope_scaling": null,
   "rope_theta": 10000.0,
   "tie_word_embeddings": false,
+  "torch_dtype": "float16",
   "transformers_version": "4.39.0",
   "use_cache": true,
   "vocab_size": 32000

model.safetensors CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2904d1e37d759d31bbdec6c0b5c226d8ca1db1bc487dd69eb3f49c75ecdcb640
-size 1242839118

 version https://git-lfs.github.com/spec/v1
+oid sha256:e50b56978024a9aed7179f00aabd8689c07912534fa25058fd873c39f3e687bd
+size 1231304860

recipe.yaml CHANGED Viewed

@@ -1,7 +1,7 @@
-quant_stage:
   quant_modifiers:
     vLLMQuantizationModifier:
-      ignore: [lm_head, model.layers.0.mlp.down_proj]
       config_groups:
         group_0:
           weights: {num_bits: 8, type: int, symmetric: true, strategy: tensor}

+test_stage:
   quant_modifiers:
     vLLMQuantizationModifier:
+      ignore: [lm_head]
       config_groups:
         group_0:
           weights: {num_bits: 8, type: int, symmetric: true, strategy: tensor}