Instructions to use TinyLlama/TinyLlama_v1.1_math_code with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use TinyLlama/TinyLlama_v1.1_math_code with Transformers:

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

pipe = pipeline("text-generation", model="TinyLlama/TinyLlama_v1.1_math_code")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("TinyLlama/TinyLlama_v1.1_math_code")
model = AutoModelForCausalLM.from_pretrained("TinyLlama/TinyLlama_v1.1_math_code", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use TinyLlama/TinyLlama_v1.1_math_code with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "TinyLlama/TinyLlama_v1.1_math_code"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "TinyLlama/TinyLlama_v1.1_math_code",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/TinyLlama/TinyLlama_v1.1_math_code

SGLang

How to use TinyLlama/TinyLlama_v1.1_math_code 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 "TinyLlama/TinyLlama_v1.1_math_code" \
    --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": "TinyLlama/TinyLlama_v1.1_math_code",
		"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 "TinyLlama/TinyLlama_v1.1_math_code" \
        --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": "TinyLlama/TinyLlama_v1.1_math_code",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use TinyLlama/TinyLlama_v1.1_math_code with Docker Model Runner:
```
docker model run hf.co/TinyLlama/TinyLlama_v1.1_math_code
```

chaoscodes commited on Jun 3, 2024

Commit

5357b80

verified ·

1 Parent(s): 59459bf

Update README.md

Browse files

Files changed (1) hide show

README.md +7 -4

README.md CHANGED Viewed

@@ -5,7 +5,10 @@ datasets:
 language:
 - en
 ---
-# TinyLlama-1.1B-v1.1 Math&Code
 https://github.com/jzhang38/TinyLlama
@@ -32,7 +35,7 @@ In this initial phase, we managed to train our model with only slimpajama to dev
 #### Continual pretraining with specific domain
-We incorporated 3 different kinds of corpus during this pretraining, slimpajama (which is the same as the first phase), Code&Math (starcoder and proof pile), and Chinese (Skypile). This approach allowed us to develop three variant models with specialized capabilities.
 At the begining ~6B tokens in this stage, we linearly increased the sampling proportion for the domain-specific corpus (excluding Slimpajama, as it remained unchanged compared with stage 1). This warmup sampling increasing strategy was designed to gradually adjust the distribution of the pretraining data, ensuring a more stable training process. After this sampling increasing stage, we continued pretraining the model with stable sampling strategy until reaching ~1.85T tokens.
@@ -45,8 +48,8 @@ Implementing a cooldown phase has become a crucial technique to achieve better m
 Following an extensive and detailed pretraining process. We are now releasing three specialized versions of our model:
 1. **TinyLlama_v1.1**: The standard version, used for general purposes.
-2. **TinyLlama_v1.1_math_code**: Equipped with better ability for math and code.
-3. **TinyLlama_v1.1_chinese**: Good understanding capacity for Chinese.
 ## Data

 language:
 - en
 ---
+<div align="center">
+# TinyLlama-1.1B-v1.1
+</div>
 https://github.com/jzhang38/TinyLlama
 #### Continual pretraining with specific domain
+We incorporated 3 different kinds of corpus during this pretraining, slimpajama (which is the same as the first phase), Math&Code (starcoder and proof pile), and Chinese (Skypile). This approach allowed us to develop three variant models with specialized capabilities.
 At the begining ~6B tokens in this stage, we linearly increased the sampling proportion for the domain-specific corpus (excluding Slimpajama, as it remained unchanged compared with stage 1). This warmup sampling increasing strategy was designed to gradually adjust the distribution of the pretraining data, ensuring a more stable training process. After this sampling increasing stage, we continued pretraining the model with stable sampling strategy until reaching ~1.85T tokens.
 Following an extensive and detailed pretraining process. We are now releasing three specialized versions of our model:
 1. **TinyLlama_v1.1**: The standard version, used for general purposes.
+2. **TinyLlama_v1.1_Math&Code**: Equipped with better ability for math and code.
+3. **TinyLlama_v1.1_Chinese**: Good understanding capacity for Chinese.
 ## Data