RedHatAI / sarvam-30b-FP8-dynamic

sarvam-30b-FP8-dynamic

Model Overview

• Model Architecture: sarvamai/sarvam-30b
  • Input: Text
  • Output: Text
• Model Optimizations:
  • Weight quantization: FP8
  • Activation quantization: FP8
• Out-of-scope: Use in any manner that violates applicable laws or regulations (including trade compliance laws).
• Version: 1.0
• Model Developers: RedHatAI

This model is a quantized version of sarvamai/sarvam-30b . It was evaluated on several tasks to assess its quality in comparison to the unquantized model.

Model Optimizations

This model was obtained by quantizing the weights and activations of sarvamai/sarvam-30b to FP8 data type, ready for inference with vLLM.

Only the weights and activations of the linear operators within transformers blocks are quantized using LLM Compressor .

Deployment

Use with vLLM

This model can be deployed efficiently using the vLLM backend.

1. Install vLLM from main:

uv pip install -U git+https://github.com/vllm-project/vllm.git \
  --extra-index-url https://wheels.vllm.ai/nightly \
  --no-deps \
  --no-cache

1. Run using vLLM

from vllm import LLM, SamplingParams
from transformers import AutoTokenizer

model_id = "RedHatAI/sarvam-30b-FP8-dynamic"
number_gpus = 1

sampling_params = SamplingParams(temperature=0.6, top_p=0.9, max_tokens=256)

tokenizer = AutoTokenizer.from_pretrained(model_id)

messages = [
    {"role": "system", "content": "You are a pirate chatbot who always responds in pirate speak!"},
    {"role": "user", "content": "Who are you?"},
]

prompts = tokenizer.apply_chat_template(messages, add_generation_prompt=True, tokenize=False)

llm = LLM(model=model_id, tensor_parallel_size=number_gpus)

outputs = llm.generate(prompts, sampling_params)

generated_text = outputs[0].outputs[0].text
print(generated_text)

vLLM also supports OpenAI-compatible serving. See the documentation for more details.

Creation

This model was created by applying LLM Compressor , as presented in the code snippet below.

uv pip install git+https://github.com/vllm-project/llm-compressor.git
uv pip install --upgrade torchvision --break-system-packages --no-cache

from compressed_tensors.offload import dispatch_model
from transformers import AutoModelForCausalLM, AutoTokenizer

from llmcompressor import oneshot
from llmcompressor.modifiers.quantization import QuantizationModifier

MODEL_ID = "sarvamai/sarvam-30b"

# Load model.
model = AutoModelForCausalLM.from_pretrained(MODEL_ID, dtype="auto", trust_remote_code=True)
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)

# Configure the quantization algorithm and scheme.
# In this case, we:
#   * quantize the weights to fp8 with per channel via ptq
#   * quantize the activations to fp8 with dynamic per token
recipe = QuantizationModifier(
    targets="Linear", scheme="FP8_DYNAMIC", ignore=["lm_head"]
)

# Apply quantization.
oneshot(model=model, recipe=recipe)

# Confirm generations of the quantized model look sane.
print("========== SAMPLE GENERATION ==============")
dispatch_model(model)
input_ids = tokenizer("Hello my name is", return_tensors="pt").input_ids.to(
    model.device
)
output = model.generate(input_ids, max_new_tokens=20)
print(tokenizer.decode(output[0]))
print("==========================================")

# Save to disk in compressed-tensors format.
SAVE_DIR = MODEL_ID.rstrip("/").split("/")[-1] + "-FP8-Dynamic"
model.save_pretrained(SAVE_DIR)
tokenizer.save_pretrained(SAVE_DIR)

Evaluation

This model was evaluated on the well-known text benchmarks using lm-evaluation-harness .

  lm_eval \
    --model vllm \
    --model_args pretrained="RedHatAI/sarvam-30b-FP8-Dynamic",dtype=auto,add_bos_token=True,max_model_len=16384,tensor_parallel_size=2,gpu_memory_utilization=0.8,enable_chunked_prefill=True,trust_remote_code=True \
    --tasks openllm \
    --write_out \
    --batch_size auto \
    --show_config

Accuracy