Instructions to use PipableAI/pip-code-bandit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use PipableAI/pip-code-bandit with Transformers:

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

pipe = pipeline("text-generation", model="PipableAI/pip-code-bandit")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("PipableAI/pip-code-bandit")
model = AutoModelForCausalLM.from_pretrained("PipableAI/pip-code-bandit")
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 PipableAI/pip-code-bandit with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/PipableAI/pip-code-bandit

SGLang

How to use PipableAI/pip-code-bandit 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 "PipableAI/pip-code-bandit" \
    --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": "PipableAI/pip-code-bandit",
		"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 "PipableAI/pip-code-bandit" \
        --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": "PipableAI/pip-code-bandit",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use PipableAI/pip-code-bandit with Docker Model Runner:
```
docker model run hf.co/PipableAI/pip-code-bandit
```

pratham-pi commited on May 13, 2024

Commit

c677009

verified ·

1 Parent(s): fb070a5

Update README.md

Browse files

Files changed (1) hide show

README.md +16 -16

README.md CHANGED Viewed

@@ -47,10 +47,9 @@ widget:
 ![image/png](https://cdn-uploads.huggingface.co/production/uploads/658d8095a2a6a6e0da8bb8a6/NuTFBTMAsPgFwMxCjdqFv.png)
-Given a goal and tools can Ai intelligently use the tools to reach the goal ?
-What if it has a meagre 1.3b params/neurons akin to that of an owl ?
-Can it follow instructions and plan to reach a goal ?
-Apparently it can.
 Releasing `pip-code-bandit` and `pipflow`
 -- a model and a library to manage and run goal oriented agentic system.
@@ -61,17 +60,18 @@ Releasing `pip-code-bandit` and `pipflow`
 -- number of params ~ 1.3b [2.9 Gb GPU memory footprint]
 -- sequence length ~ 16.3k [Can go higher but will show performance degradation]
 -- license - apache 2.0
 -- tasks:
-1. complex planning of sequential function calls with right params to accomplish a goal | a list of callables
-2. function calling | doc or code and goal
-3. code generation | plan and goal
-4. code generation | goal
-5. doc generation | code
-6. code generation | doc
-7. file recreated in json | any raw data
-8. corrected generation | new instruction with error
--- instruction following , RL tuned.
 ```
@@ -93,7 +93,7 @@ complete open sourced - apache 2.0. License
 ### NOTE:
-If you wish to try this model without utilizing your GPU, we have hosted the model on our end. To execute the library using the hosted playground model, initialize the generator as shown below:
 ```bash
 pip3 install git+https://github.com/PipableAI/pipflow.git
@@ -137,11 +137,11 @@ For detailed usage refer to the [colab_notebook](https://colab.research.google.c
 ### Model Use
 ```bash
-pip install transformers
-pip install accelerate
 ```
 ```python
 from transformers import AutoModelForCausalLM, AutoTokenizer
 from accelerate import Accelerator
 model =AutoModelForCausalLM.from_pretrained("PipableAI/pip-code-bandit",torch_dtype=torch.bfloat16,device_map="auto")

 ![image/png](https://cdn-uploads.huggingface.co/production/uploads/658d8095a2a6a6e0da8bb8a6/NuTFBTMAsPgFwMxCjdqFv.png)
+Given a goal and tools, can AI intelligently use the tools to reach the goal?
+What if it has a meagre 1.3b params/neurons akin to that of an owl? Can it follow instructions and plan to reach a goal?
+Apparently it can!
 Releasing `pip-code-bandit` and `pipflow`
 -- a model and a library to manage and run goal oriented agentic system.
 -- number of params ~ 1.3b [2.9 Gb GPU memory footprint]
 -- sequence length ~ 16.3k [Can go higher but will show performance degradation]
 -- license - apache 2.0
+-- instruction following , RL tuned.
 -- tasks:
+1. complex planning(plan) of sequential function calls | a list of callables and goal
+2. corrected plan | feedback instructions with error
+3. function calling | doc or code and goal
+4. code generation | plan and goal
+5. code generation | goal
+6. doc generation | code
+7. code generation | doc
+8. file parsed to json | any raw data
+9. sql generation | schema, question, instructions and examples
 ```
 ### NOTE:
+If you wish to try this model without utilizing your GPU, we have hosted the model on our end. To execute the library using the hosted model, initialize the generator as shown below:
 ```bash
 pip3 install git+https://github.com/PipableAI/pipflow.git
 ### Model Use
 ```bash
+pip install transformers accelerate torch
 ```
 ```python
+import torch
 from transformers import AutoModelForCausalLM, AutoTokenizer
 from accelerate import Accelerator
 model =AutoModelForCausalLM.from_pretrained("PipableAI/pip-code-bandit",torch_dtype=torch.bfloat16,device_map="auto")