⚡️ Running DeepSeek on our computer using Python

Do you want an LLM on your computer: to work offline, not leak data, and seamlessly integrate into a bot? Then let's take DeepSeek Coder and get started!

⚙️ Installing dependencies:

pip install -U transformers accelerate torch

▶️ Example code:

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_name = "deepseek-ai/deepseek-coder-6.7b-base"

tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    trust_remote_code=True,
    torch_dtype=torch.float16,   # if the GPU supports fp16
    device_map="auto"            # if there's a GPU — it will use it
)
model.eval()

prompt = "Write a Telegram feedback bot on aiogram"

inputs = tokenizer(prompt, return_tensors="pt")
device = next(model.parameters()).device
inputs = {k: v.to(device) for k, v in inputs.items()}

with torch.inference_mode():
    outputs = model.generate(
        **inputs,
        max_new_tokens=180,
        do_sample=True,      # IMPORTANT: otherwise the temperature doesn't affect
        temperature=0.7,
        top_p=0.9
    )

print(tokenizer.decode(outputs[0], skip_special_tokens=True))

➕ Advantages:
— works locally (after downloading the weights);
— easily integrates into Telegram/Discord/CLI;
— can be accelerated on the GPU via device_map="auto".

If memory is limited — there are quantized versions (4bit/8bit) and GGUF.

👍 Saving

#python #soft #code

Offline text recognition using DeepSeek-OCR AI: https://github.com/th1nhhdk/local_ai_ocr

👉 @DataScienceN

Deploy any ML model as an MCP server.

LitServe: the simplest way to roll out agents, RAG, and generally any ML models. Now with MCP support.

Completely open-source.

👉 @DataScienceN

This repository collects everything you need to use AI and LLM in your projects.

120+ libraries, organized by development stages:

→ Model training, fine-tuning, and evaluation
→ Deploying applications with LLM and RAG
→ Fast and scalable model launch
→ Data extraction, crawlers, and scrapers
→ Creating autonomous LLM agents
→ Prompt optimization and security

Repo: https://github.com/KalyanKS-NLP/llm-engineer-toolkit

🥺 https://news.1rj.ru/str/DataAnalyticsX