Instructions to use bharatgenai/AyurParam with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use bharatgenai/AyurParam with Transformers:

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

pipe = pipeline("text-generation", model="bharatgenai/AyurParam", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("bharatgenai/AyurParam", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use bharatgenai/AyurParam with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/bharatgenai/AyurParam

SGLang

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

Docker Model Runner
How to use bharatgenai/AyurParam with Docker Model Runner:
```
docker model run hf.co/bharatgenai/AyurParam
```

vLLM/SGLang Compatibility & DynamicCache Bug Report

by AA-C - opened 18 days ago

Discussion

AA-C

18 days ago

vLLM and SGLang Serving Failure (ValueError: Model architectures not supported)
Location: Repository setup / config.json

Issue: The Hugging Face repo automatically suggests serving commands such as vllm serve "bharatgenai/AyurParam" and python3 -m sglang.launch_server --model-path "bharatgenai/AyurParam".

Root Cause: Custom architectures relying on trust_remote_code=True / auto_map (ParamBharatGenForCausalLM) are not supported natively by vLLM or SGLang out of the box without defining custom model implementations or registering custom C++/CUDA kernels inside their respective codebases.

Error Triggered:

ValueError: Model architectures ['ParamBharatGenForCausalLM'] are not supported yet.
Suggested Fix: Provide instructions for overriding the model architecture locally (if compatible with Llama/Gemma), or clarify that the model requires a standard Hugging Face Transformers pipeline/FastAPI server for deployment.

Transformers 4.40+ DynamicCache Breaking Change
Location: modeling_parambharatgen.py

Issue: Generating text using recent versions of the transformers library (4.40+) fails during model inference.

Root Cause: Modern transformers generation passes DynamicCache objects for past_key_values rather than legacy tuples. The indexing and concatenation logic inside ParamBharatGenAttention in modeling_parambharatgen.py assumes past_key_values is a tuple of tensors, raising AttributeError / TypeError.

Suggested Fix: Update modeling_parambharatgen.py to support transformers.cache_utils.Cache / DynamicCache interface methods (cache_kwargs, .update()).

Unused / Backup Configuration Files
Location: tokenizer_config.json.bak, chat_template_bck.jinja

Issue: The repository contains residual .bak and _bck configuration files from development.

Suggested Fix: Clean up unused files to avoid confusion when downloading or cloning the repository programmatically.

Environment Information:
Model: bharatgenai/AyurParam

Library: transformers >= 4.40.0, vLLM >= 0.4.0

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment