Rohan Awhad RohanAwhad

Self-Hosting DeepSeek-V4.1-Flash — Hardware, Cost, Timing & 3 Acquisition Plans

Prepared for: Rohan Awhad Date: 2026-09-21 Decision: Buy (not rent) a personal, always-on inference server for DeepSeek-V4.1-Flash, with a 3–6 month procurement window.

Labeling convention used throughout

[V] = verified from a primary source (vendor page, official recipe, GitHub API, price list, arXiv/HF)
[O] = observed street/listing price at research time

SciKnowEval RL — Beginner Run Guide (default prime-rl, GRPO only)

A step-by-step guide to reproduce cold-start GRPO training on SciKnowEval with prime-rl. By the end you will have RL-trained Qwen3-8B from scratch (no SFT warmstart) on multi-domain scientific MCQ and evaluated the result.

This guide is written from an actual run (Aug 2026) on an 8×H100 node. Everything here was verified to work. Companion to the

Reverse-Text RL — Beginner Run Guide

A step-by-step guide to reproduce the reverse-text RL experiment with prime-rl. By the end you will have fine-tuned Qwen3-0.6B with RL to reverse text, starting from a warm-started (SFT) checkpoint, and evaluated the result.

This guide is written from an actual run (Aug 2026) on an 8×H100 node. Everything here was verified to work.

0. What are we actually doing?

Tensor Parallelism Knowledge Compilation

Core Understanding

Tensor Parallelism (TP) shards individual tensors across devices, following the Megatron-LM pattern:

Column-wise: Input projections (q/k/v, gate/up)
Row-wise: Output projections (o_proj, down_proj)
Sequence Parallel: Shards activations on sequence dimension for memory savings
Loss Parallel: Keep logits sharded on vocab dimension for efficient cross-entropy

	# /// script
	# dependencies = [
	# "numpy",
	# ]
	# ///

	# multi head latent attention

	import numpy as np

	#!/Users/rohan/miniconda3/bin/python

	import asyncio
	import dataclasses
	import json
	import openai
	import os
	import requests
	import sys
	import time

	import json
	import openai
	from typing import Any
	from jupyter_client.manager import KernelManager

	kernel_manager: KernelManager = None
	def execute_code_in_notebook(code: str) -> list[Any]:
	if not code: return []
	global kernel_manager
	if kernel_manager is None:

	import base64
	import cv2
	import os
	import openai
	import whisper
	import pyaudio
	import wave
	import uuid
	import threading
	from queue import Queue

	from transformers import AutoTokenizer, AutoModel
	import torch


	# MODEL CKPT is downloaded from: "jinaai/jina-embeddings-v2-base-en" # has context len of 8192
	MODEL_CKPT = "/Users/rohan/3_Resources/ai_models/jina-embeddings-v2-base-en"

	def recursive_splitter(text: str, separators: list[str], chunk_size: int) -> list[str]:
	if len(separators) == 0:
	words = text.strip().split(' ')