DeepSeek-R1-Distill-Llama-70B-quantized.w8a8

Instructions to use SAIFIINDUSTRIES/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use SAIFIINDUSTRIES/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8 with Transformers:

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

pipe = pipeline("text-generation", model="SAIFIINDUSTRIES/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("SAIFIINDUSTRIES/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8")
model = AutoModelForCausalLM.from_pretrained("SAIFIINDUSTRIES/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8", device_map="auto")
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 SAIFIINDUSTRIES/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SAIFIINDUSTRIES/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SAIFIINDUSTRIES/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/SAIFIINDUSTRIES/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8

SGLang

How to use SAIFIINDUSTRIES/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8 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 "SAIFIINDUSTRIES/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8" \
    --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": "SAIFIINDUSTRIES/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8",
		"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 "SAIFIINDUSTRIES/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8" \
        --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": "SAIFIINDUSTRIES/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use SAIFIINDUSTRIES/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8 with Docker Model Runner:
```
docker model run hf.co/SAIFIINDUSTRIES/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8
```

SAIFIINDUSTRIES

nm-research commited on 7 days ago

Commit

f8727ab

0 Parent(s):

Duplicate from RedHatAI/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8

Browse files

Co-authored-by: Neural Magic Research <nm-research@users.noreply.huggingface.co>

Files changed (24) hide show

.gitattributes +36 -0
README.md +803 -0
config.json +82 -0
generation_config.json +9 -0
model-00001-of-00015.safetensors +3 -0
model-00002-of-00015.safetensors +3 -0
model-00003-of-00015.safetensors +3 -0
model-00004-of-00015.safetensors +3 -0
model-00005-of-00015.safetensors +3 -0
model-00006-of-00015.safetensors +3 -0
model-00007-of-00015.safetensors +3 -0
model-00008-of-00015.safetensors +3 -0
model-00009-of-00015.safetensors +3 -0
model-00010-of-00015.safetensors +3 -0
model-00011-of-00015.safetensors +3 -0
model-00012-of-00015.safetensors +3 -0
model-00013-of-00015.safetensors +3 -0
model-00014-of-00015.safetensors +3 -0
model-00015-of-00015.safetensors +3 -0
model.safetensors.index.json +0 -0
recipe.yaml +21 -0
special_tokens_map.json +23 -0
tokenizer.json +3 -0
tokenizer_config.json +35 -0

.gitattributes ADDED Viewed

	@@ -0,0 +1,36 @@

+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md ADDED Viewed

	@@ -0,0 +1,803 @@

+---
+license: mit
+tags:
+- deepseek
+- int8
+- vllm
+- llmcompressor
+base_model: deepseek-ai/DeepSeek-R1-Distill-Llama-70B
+library_name: transformers
+---
+# DeepSeek-R1-Distill-Llama-70B-quantized.w8a8
+## Model Overview
+- **Model Architecture:** LlamaForCausalLM
+  - **Input:** Text
+  - **Output:** Text
+- **Model Optimizations:**
+  - **Weight quantization:** INT8
+  - **Activation quantization:** INT8
+- **Release Date:** 2/3/2025
+- **Version:** 1.0
+- **Model Developers:** Neural Magic
+Quantized version of [DeepSeek-R1-Distill-Llama-70B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-70B).
+### Model Optimizations
+This model was obtained by quantizing the weights and activations of [DeepSeek-R1-Distill-Llama-70B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Llama-70B) to INT8 data type.
+This optimization reduces the number of bits used to represent weights and activations from 16 to 8, reducing GPU memory requirements (by approximately 50%) and increasing matrix-multiply compute throughput (by approximately 2x).
+Weight quantization also reduces disk size requirements by approximately 50%.
+Only the weights and activations of the linear operators within transformers blocks are quantized.
+Weights are quantized using a symmetric per-channel scheme, whereas quantizations are quantized using a symmetric per-token scheme.
+The [GPTQ](https://arxiv.org/abs/2210.17323) algorithm is applied for quantization, as implemented in the [llm-compressor](https://github.com/vllm-project/llm-compressor) library.
+## Use with vLLM
+This model can be deployed efficiently using the [vLLM](https://docs.vllm.ai/en/latest/) backend, as shown in the example below.
+```python
+from transformers import AutoTokenizer
+from vllm import LLM, SamplingParams
+number_gpus = 2
+model_name = "neuralmagic/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+sampling_params = SamplingParams(temperature=0.6, max_tokens=256, stop_token_ids=[tokenizer.eos_token_id])
+llm = LLM(model=model_name, tensor_parallel_size=number_gpus, trust_remote_code=True)
+messages_list = [
+    [{"role": "user", "content": "Who are you? Please respond in pirate speak!"}],
+]
+prompt_token_ids = [tokenizer.apply_chat_template(messages, add_generation_prompt=True) for messages in messages_list]
+outputs = llm.generate(prompt_token_ids=prompt_token_ids, sampling_params=sampling_params)
+generated_text = [output.outputs[0].text for output in outputs]
+print(generated_text)
+```
+vLLM also supports OpenAI-compatible serving. See the [documentation](https://docs.vllm.ai/en/latest/) for more details.
+## Creation
+This model was created with [llm-compressor](https://github.com/vllm-project/llm-compressor) by running the code snippet below.
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
+from llmcompressor.modifiers.quantization import QuantizationModifier
+from llmcompressor.modifiers.smoothquant import SmoothQuantModifier
+from llmcompressor.transformers import oneshot
+from llmcompressor.transformers.compression.helpers import calculate_offload_device_map
+# Load model
+model_stub = "deepseek-ai/DeepSeek-R1-Distill-Llama-70B"
+model_name = model_stub.split("/")[-1]
+num_samples = 1024
+max_seq_len = 8192
+tokenizer = AutoTokenizer.from_pretrained(model_stub)
+device_map = calculate_offload_device_map(
+    model_stub,
+    reserve_for_hessians=True,
+    num_gpus=2,
+    torch_dtype="auto",
+)
+model = AutoModelForCausalLM.from_pretrained(
+    model_stub,
+    device_map=device_map,
+    torch_dtype="auto",
+)
+def preprocess_fn(example):
+  return {"text": tokenizer.apply_chat_template(example["messages"], add_generation_prompt=False, tokenize=False)}
+ds = load_dataset("neuralmagic/LLM_compression_calibration", split="train")
+ds = ds.map(preprocess_fn)
+# Configure the quantization algorithm and scheme
+recipe = [
+    SmoothQuantModifier(smoothing_strength=0.7),
+    QuantizationModifier(
+        targets="Linear",
+        scheme="W8A8",
+        ignore=["lm_head"],
+        dampening_frac=0.1,
+    ),
+]
+# Apply quantization
+oneshot(
+    model=model,
+    dataset=ds,
+    recipe=recipe,
+    max_seq_length=max_seq_len,
+    num_calibration_samples=num_samples,
+)
+# Save to disk in compressed-tensors format
+save_path = model_name + "-quantized.w8a8
+model.save_pretrained(save_path)
+tokenizer.save_pretrained(save_path)
+print(f"Model and tokenizer saved to: {save_path}")
+```
+## Evaluation
+The model was evaluated on OpenLLM Leaderboard [V1](https://huggingface.co/spaces/open-llm-leaderboard-old/open_llm_leaderboard) and [V2](https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard#/), using the following commands:
+OpenLLM Leaderboard V1:
+```
+lm_eval \
+  --model vllm \
+  --model_args pretrained="neuralmagic/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8",dtype=auto,max_model_len=4096,tensor_parallel_size=2,enable_chunked_prefill=True \
+  --tasks openllm \
+  --write_out \
+  --batch_size auto \
+  --output_path output_dir \
+  --show_config
+```
+OpenLLM Leaderboard V2:
+```
+lm_eval \
+  --model vllm \
+  --model_args pretrained="neuralmagic/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8",dtype=auto,max_model_len=4096,tensor_parallel_size=2,enable_chunked_prefill=True \
+  --apply_chat_template \
+  --fewshot_as_multiturn \
+  --tasks leaderboard \
+  --write_out \
+  --batch_size auto \
+  --output_path output_dir \
+  --show_config
+```
+### Accuracy
+<table>
+  <thead>
+    <tr>
+      <th>Category</th>
+      <th>Metric</th>
+      <th>deepseek-ai/DeepSeek-R1-Distill-Llama-70B</th>
+      <th>neuralmagic/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8</th>
+      <th>Recovery</th>
+    </tr>
+  </thead>
+  <tbody>
+    <tr>
+<td rowspan="4"><b>Reasoning</b></td>
+<td>AIME 2024 (pass@1)</td>
+<td>67.83</td>
+<td>67.78</td>
+<td>99.93%</td>
+</tr>
+<tr>
+<td>MATH-500 (pass@1)</td>
+<td>95.29</td>
+<td>95.27</td>
+<td>99.98%</td>
+</tr>
+<tr>
+<td>GPQA Diamond (pass@1)</td>
+<td>65.57</td>
+<td>65.01</td>
+<td>99.15%</td>
+</tr>
+<tr>
+<td><b>Average Score</b></td>
+<td><b>76.23</b></td>
+<td><b>76.02</b></td>
+<td><b>99.72%</b></td>
+</tr>
+    <tr>
+      <td rowspan="7"><b>OpenLLM V1</b></td>
+      <td>ARC-Challenge (Acc-Norm, 25-shot)</td>
+      <td>63.65</td>
+      <td>63.57</td>
+      <td>99.9%</td>
+    </tr>
+    <tr>
+      <td>GSM8K (Strict-Match, 5-shot)</td>
+      <td>93.03</td>
+      <td>93.56</td>
+      <td>100.6%</td>
+    </tr>
+    <tr>
+      <td>HellaSwag (Acc-Norm, 10-shot)</td>
+      <td>84.85</td>
+      <td>85.15</td>
+      <td>100.4%</td>
+    </tr>
+    <tr>
+      <td>MMLU (Acc, 5-shot)</td>
+      <td>78.04</td>
+      <td>78.01</td>
+      <td>100.0%</td>
+    </tr>
+    <tr>
+      <td>TruthfulQA (MC2, 0-shot)</td>
+      <td>56.67</td>
+      <td>57.47</td>
+      <td>101.4%</td>
+    </tr>
+    <tr>
+      <td>Winogrande (Acc, 5-shot)</td>
+      <td>78.22</td>
+      <td>78.37</td>
+      <td>100.2%</td>
+    </tr>
+    <tr>
+      <td><b>Average Score</b></td>
+      <td><b>75.74</b></td>
+      <td><b>76.02</b></td>
+      <td><b>100.4%</b></td>
+    </tr>
+    <tr>
+      <td rowspan="7"><b>OpenLLM V2</b></td>
+      <td>IFEval (Inst Level Strict Acc, 0-shot)</td>
+      <td>42.45</td>
+      <td>42.51</td>
+      <td>100.1%</td>
+    </tr>
+    <tr>
+      <td>BBH (Acc-Norm, 3-shot)</td>
+      <td>21.26</td>
+      <td>20.78</td>
+      <td>97.8%</td>
+    </tr>
+    <tr>
+      <td>Math-Hard (Exact-Match, 4-shot)</td>
+      <td>0.00</td>
+      <td>0.00</td>
+      <td>---</td>
+    </tr>
+    <tr>
+      <td>GPQA (Acc-Norm, 0-shot)</td>
+      <td>9.51</td>
+      <td>7.25</td>
+      <td>---</td>
+    </tr>
+    <tr>
+      <td>MUSR (Acc-Norm, 0-shot)</td>
+      <td>14.87</td>
+      <td>15.24</td>
+      <td>---</td>
+    </tr>
+    <tr>
+      <td>MMLU-Pro (Acc, 5-shot)</td>
+      <td>4.27</td>
+      <td>5.62</td>
+      <td>---</td>
+    </tr>
+    <tr>
+      <td><b>Average Score</b></td>
+      <td><b>15.39</b></td>
+      <td><b>15.23</b></td>
+      <td><b>99.0%</b></td>
+    </tr>
+    <tr>
+      <td rowspan="4"><b>Coding</b></td>
+      <td>HumanEval (pass@1)</td>
+      <td>81.10</td>
+      <td>81.00</td>
+      <td><b>99.9%</b></td>
+    </tr>
+    <tr>
+      <td>HumanEval (pass@10)</td>
+      <td>87.60</td>
+      <td>86.80</td>
+      <td>99.1%</td>
+    </tr>
+    <tr>
+      <td>HumanEval+ (pass@10)</td>
+      <td>75.20</td>
+      <td>75.80</td>
+      <td>100.8%</td>
+    </tr>
+    <tr>
+      <td>HumanEval+ (pass@10)</td>
+      <td>83.10</td>
+      <td>83.40</td>
+      <td>100.4%</td>
+    </tr>
+  </tbody>
+</table>
+## Inference Performance
+This model achieves up to 2.0x speedup in single-stream deployment and up to 2.2x speedup in multi-stream asynchronous deployment, depending on hardware and use-case scenario.
+The following performance benchmarks were conducted with [vLLM](https://docs.vllm.ai/en/latest/) version 0.7.2, and [GuideLLM](https://github.com/neuralmagic/guidellm).
+<details>
+<summary>Benchmarking Command</summary>
+```
+guidellm --model neuralmagic/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8 --target "http://localhost:8000/v1" --data-type emulated --data "prompt_tokens=<prompt_tokens>,generated_tokens=<generated_tokens>" --max seconds 360 --backend aiohttp_server
+```
+</details>
+### Single-stream performance (measured with vLLM version 0.7.2)
+<table>
+  <thead>
+    <tr>
+      <th></th>
+      <th></th>
+      <th></th>
+      <th></th>
+      <th style="text-align: center;" colspan="2" >Instruction Following<br>256 / 128</th>
+      <th style="text-align: center;" colspan="2" >Multi-turn Chat<br>512 / 256</th>
+      <th style="text-align: center;" colspan="2" >Docstring Generation<br>768 / 128</th>
+      <th style="text-align: center;" colspan="2" >RAG<br>1024 / 128</th>
+      <th style="text-align: center;" colspan="2" >Code Completion<br>256 / 1024</th>
+      <th style="text-align: center;" colspan="2" >Code Fixing<br>1024 / 1024</th>
+      <th style="text-align: center;" colspan="2" >Large Summarization<br>4096 / 512</th>
+      <th style="text-align: center;" colspan="2" >Large RAG<br>10240 / 1536</th>
+    </tr>
+    <tr>
+      <th>GPU class</th>
+      <th>Number of GPUs</th>
+      <th>Model</th>
+      <th>Average cost reduction</th>
+      <th>Latency (s)</th>
+      <th>QPD</th>
+      <th>Latency (s)</th>
+      <th>QPD</th>
+      <th>Latency (s)</th>
+      <th>QPD</th>
+      <th>Latency (s)</th>
+      <th>QPD</th>
+      <th>Latency (s)</th>
+      <th>QPD</th>
+      <th>Latency (s)</th>
+      <th>QPD</th>
+      <th>Latency (s)</th>
+      <th>QPD</th>
+      <th>Latency (s)</th>
+      <th>QPD</th>
+    </tr>
+  </thead>
+  <tbody style="text-align: center" >
+    <tr>
+      <th rowspan="3" valign="top">A6000</th>
+      <td>4</td>
+      <th>deepseek-ai/DeepSeek-R1-Distill-Llama-70B</th>
+      <td>---</td>
+      <td>7.4</td>
+      <td>152</td>
+      <td>14.9</td>
+      <td>76</td>
+      <td>7.5</td>
+      <td>149</td>
+      <td>7.7</td>
+      <td>146</td>
+      <td>57.2</td>
+      <td>20</td>
+      <td>58.9</td>
+      <td>19</td>
+      <td>31.9</td>
+      <td>35</td>
+      <td>98.4</td>
+      <td>11</td>
+    </tr>
+    <tr>
+      <td>2</td>
+      <th>neuralmagic/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8</th>
+      <td>1.93</td>
+      <td>7.7</td>
+      <td>292</td>
+      <td>15.2</td>
+      <td>148</td>
+      <td>7.8</td>
+      <td>287</td>
+      <td>8.0</td>
+      <td>282</td>
+      <td>60.7</td>
+      <td>37</td>
+      <td>60.2</td>
+      <td>37</td>
+      <td>32.3</td>
+      <td>70</td>
+      <td>104.0</td>
+      <td>22</td>
+    </tr>
+    <tr>
+      <td>2</td>
+      <th>neuralmagic/DeepSeek-R1-Distill-Llama-70B-quantized.w4a16</th>
+      <td>2.83</td>
+      <td>4.9</td>
+      <td>457</td>
+      <td>10.0</td>
+      <td>225</td>
+      <td>5.5</td>
+      <td>411</td>
+      <td>5.8</td>
+      <td>389</td>
+      <td>38.9</td>
+      <td>58</td>
+      <td>39.2</td>
+      <td>57</td>
+      <td>23.7</td>
+      <td>95</td>
+      <td>76.6</td>
+      <td>29</td>
+    </tr>
+    <tr>
+      <th rowspan="3" valign="top">A100</th>
+      <td>2</td>
+      <th>deepseek-ai/DeepSeek-R1-Distill-Llama-70B</th>
+      <td>---</td>
+      <td>6.4</td>
+      <td>157</td>
+      <td>12.8</td>
+      <td>79</td>
+      <td>6.6</td>
+      <td>153</td>
+      <td>6.7</td>
+      <td>151</td>
+      <td>50.4</td>
+      <td>20</td>
+      <td>50.8</td>
+      <td>20</td>
+      <td>27.0</td>
+      <td>37</td>
+      <td>85.4</td>
+      <td>12</td>
+    </tr>
+    <tr>
+      <td>2</td>
+      <th>neuralmagic/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8</th>
+      <td>1.48</td>
+      <td>4.1</td>
+      <td>245</td>
+      <td>8.2</td>
+      <td>123</td>
+      <td>4.2</td>
+      <td>238</td>
+      <td>4.3</td>
+      <td>235</td>
+      <td>32.4</td>
+      <td>31</td>
+      <td>32.8</td>
+      <td>31</td>
+      <td>17.6</td>
+      <td>57</td>
+      <td>90.8</td>
+      <td>11</td>
+    </tr>
+    <tr>
+      <td>1</td>
+      <th>neuralmagic/DeepSeek-R1-Distill-Llama-70B-quantized.w4a16</th>
+      <td>2.69</td>
+      <td>4.6</td>
+      <td>440</td>
+      <td>9.2</td>
+      <td>220</td>
+      <td>4.9</td>
+      <td>407</td>
+      <td>5.2</td>
+      <td>389</td>
+      <td>35.3</td>
+      <td>57</td>
+      <td>36.3</td>
+      <td>55</td>
+      <td>21.2</td>
+      <td>95</td>
+      <td>68.1</td>
+      <td>30</td>
+    </tr>
+    <tr>
+      <th rowspan="3" valign="top">H100</th>
+      <td>2</td>
+      <th>deepseek-ai/DeepSeek-R1-Distill-Llama-70B</th>
+      <td>---</td>
+      <td>3.8</td>
+      <td>149</td>
+      <td>7.6</td>
+      <td>74</td>
+      <td>3.9</td>
+      <td>146</td>
+      <td>3.9</td>
+      <td>144</td>
+      <td>30.0</td>
+      <td>19</td>
+      <td>30.4</td>
+      <td>19</td>
+      <td>16.1</td>
+      <td>35</td>
+      <td>56.5</td>
+      <td>10</td>
+    </tr>
+    <tr>
+      <td>2</td>
+      <th>neuralmagic/DeepSeek-R1-Distill-Llama-70B-FP8-dynamic</th>
+      <td>1.39</td>
+      <td>2.7</td>
+      <td>210</td>
+      <td>5.3</td>
+      <td>106</td>
+      <td>2.7</td>
+      <td>207</td>
+      <td>2.8</td>
+      <td>203</td>
+      <td>21.1</td>
+      <td>27</td>
+      <td>21.4</td>
+      <td>26</td>
+      <td>11.5</td>
+      <td>49</td>
+      <td>47.2</td>
+      <td>12</td>
+    </tr>
+    <tr>
+      <td>1</td>
+      <th>neuralmagic/DeepSeek-R1-Distill-Llama-70B-quantized.w4a16</th>
+      <td>1.83</td>
+      <td>4.0</td>
+      <td>277</td>
+      <td>7.9</td>
+      <td>138</td>
+      <td>4.1</td>
+      <td>266</td>
+      <td>4.2</td>
+      <td>262</td>
+      <td>31.2</td>
+      <td>35</td>
+      <td>31.8</td>
+      <td>34</td>
+      <td>17.8</td>
+      <td>61</td>
+      <td>61.4</td>
+      <td>18</td>
+    </tr>
+  </tbody>
+</table>
+**Use case profiles: prompt tokens / generation tokens
+**QPD: Queries per dollar, based on on-demand cost at [Lambda Labs](https://lambdalabs.com/service/gpu-cloud) (observed on 2/18/2025).
+### Multi-stream asynchronous performance (measured with vLLM version 0.7.2)
+<table>
+  <thead>
+    <tr>
+      <th></th>
+      <th></th>
+      <th></th>
+      <th style="text-align: center;" colspan="2" >Instruction Following<br>256 / 128</th>
+      <th style="text-align: center;" colspan="2" >Multi-turn Chat<br>512 / 256</th>
+      <th style="text-align: center;" colspan="2" >Docstring Generation<br>768 / 128</th>
+      <th style="text-align: center;" colspan="2" >RAG<br>1024 / 128</th>
+      <th style="text-align: center;" colspan="2" >Code Completion<br>256 / 1024</th>
+      <th style="text-align: center;" colspan="2" >Code Fixing<br>1024 / 1024</th>
+      <th style="text-align: center;" colspan="2" >Large Summarization<br>4096 / 512</th>
+      <th style="text-align: center;" colspan="2" >Large RAG<br>10240 / 1536</th>
+    </tr>
+    <tr>
+      <th>Hardware</th>
+      <th>Model</th>
+      <th>Average cost reduction</th>
+      <th>Maximum throughput (QPS)</th>
+      <th>QPD</th>
+      <th>Maximum throughput (QPS)</th>
+      <th>QPD</th>
+      <th>Maximum throughput (QPS)</th>
+      <th>QPD</th>
+      <th>Maximum throughput (QPS)</th>
+      <th>QPD</th>
+      <th>Maximum throughput (QPS)</th>
+      <th>QPD</th>
+      <th>Maximum throughput (QPS)</th>
+      <th>QPD</th>
+      <th>Maximum throughput (QPS)</th>
+      <th>QPD</th>
+      <th>Maximum throughput (QPS)</th>
+      <th>QPD</th>
+    </tr>
+  </thead>
+  <tbody style="text-align: center" >
+    <tr>
+      <th rowspan="3" valign="top">A6000x4</th>
+      <th>deepseek-ai/DeepSeek-R1-Distill-Llama-70B</th>
+      <td>---</td>
+      <td>3.65</td>
+      <td>4102</td>
+      <td>1.56</td>
+      <td>1757</td>
+      <td>1.90</td>
+      <td>2143</td>
+      <td>1.48</td>
+      <td>1665</td>
+      <td>0.44</td>
+      <td>493</td>
+      <td>0.34</td>
+      <td>380</td>
+      <td>0.22</td>
+      <td>245</td>
+      <td>0.05</td>
+      <td>55</td>
+    </tr>
+    <tr>
+      <th>neuralmagic/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8</th>
+      <td>1.76</td>
+      <td>5.89</td>
+      <td>6625</td>
+      <td>2.94</td>
+      <td>3307</td>
+      <td>3.36</td>
+      <td>3775</td>
+      <td>2.59</td>
+      <td>2916</td>
+      <td>0.74</td>
+      <td>828</td>
+      <td>0.53</td>
+      <td>601</td>
+      <td>0.35</td>
+      <td>398</td>
+      <td>0.11</td>
+      <td>120</td>
+    </tr>
+    <tr>
+      <th>neuralmagic/DeepSeek-R1-Distill-Llama-70B-quantized.w4a16</th>
+      <td>1.48</td>
+      <td>4.91</td>
+      <td>5528</td>
+      <td>2.01</td>
+      <td>2259</td>
+      <td>2.03</td>
+      <td>2280</td>
+      <td>1.12</td>
+      <td>1255</td>
+      <td>1.11</td>
+      <td>1251</td>
+      <td>0.76</td>
+      <td>852</td>
+      <td>0.24</td>
+      <td>267</td>
+      <td>0.07</td>
+      <td>81</td>
+    </tr>
+    <tr>
+      <th rowspan="3" valign="top">A100x4</th>
+      <th>deepseek-ai/DeepSeek-R1-Distill-Llama-70B</th>
+      <td>---</td>
+      <td>10.41</td>
+      <td>5235</td>
+      <td>5.10</td>
+      <td>2565</td>
+      <td>5.50</td>
+      <td>2766</td>
+      <td>4.36</td>
+      <td>2193</td>
+      <td>1.49</td>
+      <td>751</td>
+      <td>1.21</td>
+      <td>607</td>
+      <td>0.89</td>
+      <td>447</td>
+      <td>0.19</td>
+      <td>98</td>
+    </tr>
+    <tr>
+      <th>neuralmagic/DeepSeek-R1-Distill-Llama-70B-quantized.w8a8</th>
+      <td>1.63</td>
+      <td>18.11</td>
+      <td>9103</td>
+      <td>8.90</td>
+      <td>4477</td>
+      <td>9.41</td>
+      <td>4730</td>
+      <td>7.42</td>
+      <td>3731</td>
+      <td>2.44</td>
+      <td>1229</td>
+      <td>1.89</td>
+      <td>948</td>
+      <td>1.26</td>
+      <td>631</td>
+      <td>0.30</td>
+      <td>149</td>
+    </tr>
+    <tr>
+      <th>neuralmagic/DeepSeek-R1-Distill-Llama-70B-quantized.w4a16</th>
+      <td>1.12</td>
+      <td>12.63</td>
+      <td>6353</td>
+      <td>5.32</td>
+      <td>2673</td>
+      <td>5.58</td>
+      <td>2804</td>
+      <td>4.27</td>
+      <td>2144</td>
+      <td>2.30</td>
+      <td>1158</td>
+      <td>1.45</td>
+      <td>729</td>
+      <td>0.76</td>
+      <td>381</td>
+      <td>0.22</td>
+      <td>110</td>
+    </tr>
+    <tr>
+      <th rowspan="3" valign="top">H100x4</th>
+      <th>deepseek-ai/DeepSeek-R1-Distill-Llama-70B</th>
+      <td>---</td>
+      <td>14.04</td>
+      <td>2113</td>
+      <td>10.85</td>
+      <td>1634</td>
+      <td>12.25</td>
+      <td>1844</td>
+      <td>9.93</td>
+      <td>1494</td>
+      <td>3.68</td>
+      <td>554</td>
+      <td>2.82</td>
+      <td>425</td>
+      <td>1.81</td>
+      <td>273</td>
+      <td>0.35</td>
+      <td>52</td>
+    </tr>
+    <tr>
+      <th>neuralmagic/DeepSeek-R1-Distill-Llama-70B-FP8-dynamic</th>
+      <td>1.78</td>
+      <td>41.44</td>
+      <td>6236</td>
+      <td>19.64</td>
+      <td>2956</td>
+      <td>21.03</td>
+      <td>3166</td>
+      <td>16.72</td>
+      <td>2516</td>
+      <td>6.01</td>
+      <td>904</td>
+      <td>4.46</td>
+      <td>672</td>
+      <td>2.55</td>
+      <td>383</td>
+      <td>0.49</td>
+      <td>74</td>
+    </tr>
+    <tr>
+      <th>neuralmagic/DeepSeek-R1-Distill-Llama-70B-quantized.w4a16</th>
+      <td>1.45</td>
+      <td>36.61</td>
+      <td>5509</td>
+      <td>15.12</td>
+      <td>2275</td>
+      <td>16.24</td>
+      <td>2443</td>
+      <td>13.22</td>
+      <td>1990</td>
+      <td>5.48</td>
+      <td>825</td>
+      <td>3.01</td>
+      <td>453</td>
+      <td>2.07</td>
+      <td>312</td>
+      <td>0.43</td>
+      <td>64</td>
+    </tr>
+  </tbody>
+</table>
+**Use case profiles: prompt tokens / generation tokens
+**QPS: Queries per second.
+**QPD: Queries per dollar, based on on-demand cost at [Lambda Labs](https://lambdalabs.com/service/gpu-cloud) (observed on 2/18/2025).

config.json ADDED Viewed

	@@ -0,0 +1,82 @@

+{
+  "_name_or_path": "deepseek-ai/DeepSeek-R1-Distill-Llama-70B",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": [
+    128001,
+    128008,
+    128009
+  ],
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 8192,
+  "initializer_range": 0.02,
+  "intermediate_size": 28672,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 64,
+  "num_hidden_layers": 80,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "quantization_config": {
+    "config_groups": {
+      "group_0": {
+        "input_activations": {
+          "actorder": null,
+          "block_structure": null,
+          "dynamic": true,
+          "group_size": null,
+          "num_bits": 8,
+          "observer": null,
+          "observer_kwargs": {},
+          "strategy": "token",
+          "symmetric": true,
+          "type": "int"
+        },
+        "output_activations": null,
+        "targets": [
+          "Linear"
+        ],
+        "weights": {
+          "actorder": null,
+          "block_structure": null,
+          "dynamic": false,
+          "group_size": null,
+          "num_bits": 8,
+          "observer": "mse",
+          "observer_kwargs": {},
+          "strategy": "channel",
+          "symmetric": true,
+          "type": "int"
+        }
+      }
+    },
+    "format": "int-quantized",
+    "global_compression_ratio": 1.5343121209820911,
+    "ignore": [
+      "lm_head"
+    ],
+    "kv_cache_scheme": null,
+    "quant_method": "compressed-tensors",
+    "quantization_status": "compressed"
+  },
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.48.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}

generation_config.json ADDED Viewed

	@@ -0,0 +1,9 @@

+{
+  "_from_model_config": true,
+  "bos_token_id": 128000,
+  "do_sample": true,
+  "eos_token_id": 128001,
+  "temperature": 0.6,
+  "top_p": 0.95,
+  "transformers_version": "4.48.0"
+}

model-00001-of-00015.safetensors ADDED Viewed

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

+version https://git-lfs.github.com/spec/v1
+oid sha256:874441b9d160380a02a42c15fd9a1b8d0b665140cfe0cd547f8a07d5fb655573
+size 4819900928

model-00002-of-00015.safetensors ADDED Viewed

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

+version https://git-lfs.github.com/spec/v1
+oid sha256:4d09ea24018a9460f254adf6bb743c31e8419bf40a9a46ae1a073924205a84d1
+size 4984010360

model-00003-of-00015.safetensors ADDED Viewed

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

+version https://git-lfs.github.com/spec/v1
+oid sha256:79754d2c832a36e4eaeae7d3e4470159659e0a97345237e9d065d528b40d9beb
+size 4900112536

model-00004-of-00015.safetensors ADDED Viewed

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

+version https://git-lfs.github.com/spec/v1
+oid sha256:2fc0a77e20010ba725d9f32815d067b9b76b352238689a817c02ce1ba8c744f2
+size 4900104584

model-00005-of-00015.safetensors ADDED Viewed

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

+version https://git-lfs.github.com/spec/v1
+oid sha256:9624b46d340f8cc99a797c1b7b1182b1547a020ec028748020ddfff3d077c08b
+size 4900104584

model-00006-of-00015.safetensors ADDED Viewed

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

+version https://git-lfs.github.com/spec/v1
+oid sha256:fec67f8dafcaf3953370cbff0398c61c1028bbf8b5d4f4629f64c4006e7d5a9d
+size 4984010448

model-00007-of-00015.safetensors ADDED Viewed

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

+version https://git-lfs.github.com/spec/v1
+oid sha256:8c025d3c4ca322be0e58b10f5a002d7cb79a20ee92a265e98ad8c57f2a83eaaf
+size 4900112552

model-00008-of-00015.safetensors ADDED Viewed

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

+version https://git-lfs.github.com/spec/v1
+oid sha256:63aff9cefddaa23394dbc25d6c8620950c970f4887ee901b2c155ad6cda9e1b0
+size 4900104584

model-00009-of-00015.safetensors ADDED Viewed

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

+version https://git-lfs.github.com/spec/v1
+oid sha256:e5b118bd996dc131d567b468c7d591b9bc62904d5b15c6e408dabccebfad5004
+size 4900104584

model-00010-of-00015.safetensors ADDED Viewed

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

+version https://git-lfs.github.com/spec/v1
+oid sha256:8b68bfba415890a0b2532be3c5dea1022ff860758005560dd518fb3d20a4c985
+size 4984010448

model-00011-of-00015.safetensors ADDED Viewed

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

+version https://git-lfs.github.com/spec/v1
+oid sha256:2209aa58b5a764845b76754d7b59286a44afcaa402ff783ea736a62760dcf166
+size 4900112552

model-00012-of-00015.safetensors ADDED Viewed

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

+version https://git-lfs.github.com/spec/v1
+oid sha256:caf0452804b710de11de3cfa5b9dfafe20ee0e494ca7bf12623232482f44a383
+size 4900104584

model-00013-of-00015.safetensors ADDED Viewed

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

+version https://git-lfs.github.com/spec/v1
+oid sha256:52adbf1b59575136df72166bafd03700f1b97bc84c2e98de660f4b7409d13f03
+size 4900104584

model-00014-of-00015.safetensors ADDED Viewed

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

+version https://git-lfs.github.com/spec/v1
+oid sha256:47f34fecb33e7d880fbe71b57f7b863e1aacc1941810f2f3694a68839b3a2e3b
+size 4984010448

model-00015-of-00015.safetensors ADDED Viewed

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

+version https://git-lfs.github.com/spec/v1
+oid sha256:dffb1c41e7f51c021a80afe82e0baf5165cc3642b980281510c510d5c7b6e311
+size 3813044096

model.safetensors.index.json ADDED Viewed

The diff for this file is too large to render. See raw diff

recipe.yaml ADDED Viewed

	@@ -0,0 +1,21 @@

+quant_stage:
+  quant_modifiers:
+    SmoothQuantModifier:
+      smoothing_strength: 0.7
+      mappings:
+      - - ['re:.*q_proj', 're:.*k_proj', 're:.*v_proj']
+        - re:.*input_layernorm
+      - - ['re:.*gate_proj', 're:.*up_proj']
+        - re:.*post_attention_layernorm
+      - - ['re:.*down_proj']
+        - re:.*up_proj
+    GPTQModifier:
+      sequential_update: true
+      dampening_frac: 0.1
+      ignore: [lm_head]
+      config_groups:
+        group_0:
+          targets: [Linear]
+          weights: {num_bits: 8, type: int, symmetric: true, strategy: channel, observer: mse}
+          input_activations: {num_bits: 8, type: int, symmetric: true, strategy: token, dynamic: true,
+            observer: memoryless}

special_tokens_map.json ADDED Viewed

	@@ -0,0 +1,23 @@

+{
+  "bos_token": {
+    "content": "<｜begin▁of▁sentence｜>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<｜end▁of▁sentence｜>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<｜end▁of▁sentence｜>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.json ADDED Viewed

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

+version https://git-lfs.github.com/spec/v1
+oid sha256:b9c9eb63a8e03059914880f918cd28a880dec8b6e15e4461e1ff677e3743dbb8
+size 9084480

tokenizer_config.json ADDED Viewed

	@@ -0,0 +1,35 @@

+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "bos_token": {
+    "__type": "AddedToken",
+    "content": "<｜begin▁of▁sentence｜>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "clean_up_tokenization_spaces": false,
+  "eos_token": {
+    "__type": "AddedToken",
+    "content": "<｜end▁of▁sentence｜>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "legacy": true,
+  "model_max_length": 16384,
+  "pad_token": {
+    "__type": "AddedToken",
+    "content": "<｜end▁of▁sentence｜>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "sp_model_kwargs": {},
+  "unk_token": null,
+  "tokenizer_class": "LlamaTokenizerFast",
+  "chat_template": "{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% set ns = namespace(is_first=false, is_tool=false, is_output_first=true, system_prompt='') %}{%- for message in messages %}{%- if message['role'] == 'system' %}{% set ns.system_prompt = message['content'] %}{%- endif %}{%- endfor %}{{bos_token}}{{ns.system_prompt}}{%- for message in messages %}{%- if message['role'] == 'user' %}{%- set ns.is_tool = false -%}{{'<｜User｜>' + message['content']}}{%- endif %}{%- if message['role'] == 'assistant' and message['content'] is none %}{%- set ns.is_tool = false -%}{%- for tool in message['tool_calls']%}{%- if not ns.is_first %}{{'<｜Assistant｜><｜tool▁calls▁begin｜><｜tool▁call▁begin｜>' + tool['type'] + '<｜tool▁sep｜>' + tool['function']['name'] + '\\n' + '```json' + '\\n' + tool['function']['arguments'] + '\\n' + '```' + '<｜tool▁call▁end｜>'}}{%- set ns.is_first = true -%}{%- else %}{{'\\n' + '<｜tool▁call▁begin｜>' + tool['type'] + '<｜tool▁sep｜>' + tool['function']['name'] + '\\n' + '```json' + '\\n' + tool['function']['arguments'] + '\\n' + '```' + '<｜tool▁call▁end｜>'}}{{'<｜tool▁calls▁end｜><｜end▁of▁sentence｜>'}}{%- endif %}{%- endfor %}{%- endif %}{%- if message['role'] == 'assistant' and message['content'] is not none %}{%- if ns.is_tool %}{{'<｜tool▁outputs▁end｜>' + message['content'] + '<｜end▁of▁sentence｜>'}}{%- set ns.is_tool = false -%}{%- else %}{% set content = message['content'] %}{% if '</think>' in content %}{% set content = content.split('</think>')[-1] %}{% endif %}{{'<｜Assistant｜>' + content + '<｜end▁of▁sentence｜>'}}{%- endif %}{%- endif %}{%- if message['role'] == 'tool' %}{%- set ns.is_tool = true -%}{%- if ns.is_output_first %}{{'<｜tool▁outputs▁begin｜><｜tool▁output▁begin｜>' + message['content'] + '<｜tool▁output▁end｜>'}}{%- set ns.is_output_first = false %}{%- else %}{{'\\n<｜tool▁output▁begin｜>' + message['content'] + '<｜tool▁output▁end｜>'}}{%- endif %}{%- endif %}{%- endfor -%}{% if ns.is_tool %}{{'<｜tool▁outputs▁end｜>'}}{% endif %}{% if add_generation_prompt and not ns.is_tool %}{{'<｜Assistant｜><think>\\n'}}{% endif %}"
+}