Added Cerebras LLM for ~2000 TPS inference

[sk5268]

Purpose

Cerebras using its 3rd generation Wafer-Scale-Engine (WSE-3) can make inference at ~2000 TPS. So I added Cerebras API via cerebras-cloud-sdk

Usage

import os
from beyondllm.llms.cerebras import CerebrasModel

os.environ['CEREBRAS_API_KEY'] = "YOUR_CEREBRAS_API_KEY"

llm = CerebrasModel()

# Generate a prediction
prompt = "Tell me a about machine learning"
response = llm.predict(prompt)
print(response)

[sk5268]

@lucifertrj