Cascade-Droidz

A fine-tuned version of nvidia/Nemotron-Cascade-14B-Thinking optimized for chat applications.

Training

This model uses an experimental approach to fine-tuning with a custom dataset designed for conversational tasks.

Inference

Transformers

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained(
    "Daemontatox/Cascade-Droidz",
    device_map="auto",
    torch_dtype="auto"
)
tokenizer = AutoTokenizer.from_pretrained("Daemontatox/Cascade-Droidz")

messages = [
    {"role": "user", "content": "Hello, how are you?"}
]

input_ids = tokenizer.apply_chat_template(
    messages,
    add_generation_prompt=True,
    return_tensors="pt"
).to(model.device)

outputs = model.generate(
    input_ids,
    max_new_tokens=512,
    temperature=0.7,
    top_p=0.9,
    do_sample=True
)

response = tokenizer.decode(outputs[0][input_ids.shape[-1]:], skip_special_tokens=True)
print(response)

vLLM

from vllm import LLM, SamplingParams

llm = LLM(
    model="Daemontatox/Cascade-Droidz",
    tensor_parallel_size=4,  # Adjust based on GPU count
    dtype="auto"
)

sampling_params = SamplingParams(
    temperature=0.7,
    top_p=0.9,
    max_tokens=512
)

prompts = ["Hello, how are you?"]
outputs = llm.generate(prompts, sampling_params)

for output in outputs:
    print(output.outputs[0].text)

vLLM OpenAI-Compatible Server

vllm serve Daemontatox/Cascade-Droidz \
    --tensor-parallel-size 4 \
    --dtype auto \
    --max-model-len 4096

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:8000/v1",
    api_key="token-abc123"
)

response = client.chat.completions.create(
    model="Daemontatox/Cascade-Droidz",
    messages=[
        {"role": "user", "content": "Hello, how are you?"}
    ],
    temperature=0.7,
    max_tokens=512
)

print(response.choices[0].message.content)

TensorRT-LLM

Convert to TensorRT-LLM format

convert_checkpoint.py \
    --model_dir Daemontatox/Cascade-Droidz \
    --output_dir ./trt_ckpt \
    --dtype float16 \
    --tp_size 4

Build TensorRT engine

trtllm-build \
    --checkpoint_dir ./trt_ckpt \
    --output_dir ./trt_engine \
    --gemm_plugin float16 \
    --max_batch_size 8 \
    --max_input_len 2048 \
    --max_output_len 512

from tensorrt_llm import LLM

llm = LLM(model="./trt_engine")

prompts = ["Hello, how are you?"]
outputs = llm.generate(prompts, max_new_tokens=512)

for output in outputs:
    print(output.text)

Modular MAX

# Serve with MAX Engine
max serve Daemontatox/Cascade-Droidz \
    --port 8000 \
    --tensor-parallel-size 4

from max import engine

# Load model with MAX
model = engine.InferenceSession(
    "Daemontatox/Cascade-Droidz",
    device="cuda",
    tensor_parallel=4
)

# Run inference
prompt = "Hello, how are you?"
output = model.generate(
    prompt,
    max_tokens=512,
    temperature=0.7,
    top_p=0.9
)

print(output.text)

# Using MAX with Python API
from max.serve import serve
from max.pipelines import pipeline

# Create pipeline
pipe = pipeline(
    "text-generation",
    model="Daemontatox/Cascade-Droidz",
    device="cuda",
    tensor_parallel=4
)

# Generate
result = pipe(
    "Hello, how are you?",
    max_new_tokens=512,
    temperature=0.7,
    top_p=0.9
)

print(result[0]["generated_text"])

Limitations

As this uses experimental training methods, results may vary. Test thoroughly before production use. ```