Instructions to use trillionlabs/Tri-70B-preview-SFT with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use trillionlabs/Tri-70B-preview-SFT with Transformers:

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

pipe = pipeline("text-generation", model="trillionlabs/Tri-70B-preview-SFT", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("trillionlabs/Tri-70B-preview-SFT", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use trillionlabs/Tri-70B-preview-SFT with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/trillionlabs/Tri-70B-preview-SFT

SGLang

How to use trillionlabs/Tri-70B-preview-SFT 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 "trillionlabs/Tri-70B-preview-SFT" \
    --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": "trillionlabs/Tri-70B-preview-SFT",
		"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 "trillionlabs/Tri-70B-preview-SFT" \
        --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": "trillionlabs/Tri-70B-preview-SFT",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use trillionlabs/Tri-70B-preview-SFT with Docker Model Runner:
```
docker model run hf.co/trillionlabs/Tri-70B-preview-SFT
```

WonsukYangTL commited on Jan 8

Commit

236a48f

verified ·

1 Parent(s): e57c578

Update configuration_trillion.py

Browse files

Files changed (1) hide show

configuration_trillion.py +6 -0

configuration_trillion.py CHANGED Viewed

@@ -174,6 +174,7 @@ class TrillionConfig(PretrainedConfig):
         rope_scaling=None,
         global_attention_freq=4,
         attn_temperature_tuning=True,
         floor_scale=1.0,
         attn_scale=1.0,
         attention_bias=False,
@@ -183,6 +184,7 @@ class TrillionConfig(PretrainedConfig):
         **kwargs,
     ):
         self.attn_temperature_tuning = attn_temperature_tuning
         self.attn_scale = attn_scale
         self.floor_scale = floor_scale
         self.global_attention_freq = global_attention_freq
@@ -215,6 +217,10 @@ class TrillionConfig(PretrainedConfig):
             self.rope_scaling["rope_type"] = self.rope_scaling["type"]
         rope_config_validation(self)
         super().__init__(
             pad_token_id=pad_token_id,
             bos_token_id=bos_token_id,

         rope_scaling=None,
         global_attention_freq=4,
         attn_temperature_tuning=True,
+        sliding_window=4096,
         floor_scale=1.0,
         attn_scale=1.0,
         attention_bias=False,
         **kwargs,
     ):
         self.attn_temperature_tuning = attn_temperature_tuning
+        self.sliding_window = sliding_window
         self.attn_scale = attn_scale
         self.floor_scale = floor_scale
         self.global_attention_freq = global_attention_freq
             self.rope_scaling["rope_type"] = self.rope_scaling["type"]
         rope_config_validation(self)
+        self.layer_types = [
+            "sliding_attention" if bool((i + 1) % self.global_attention_freq) else "full_attention" for i in range(self.num_hidden_layers)
+        ]
         super().__init__(
             pad_token_id=pad_token_id,
             bos_token_id=bos_token_id,