Instructions to use AGmind/agmind-rag-splitter-ru with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use AGmind/agmind-rag-splitter-ru with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="AGmind/agmind-rag-splitter-ru",
	filename="splitter-ru-8b-Q5_K_M.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use AGmind/agmind-rag-splitter-ru with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf AGmind/agmind-rag-splitter-ru:Q5_K_M
# Run inference directly in the terminal:
llama cli -hf AGmind/agmind-rag-splitter-ru:Q5_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf AGmind/agmind-rag-splitter-ru:Q5_K_M
# Run inference directly in the terminal:
llama cli -hf AGmind/agmind-rag-splitter-ru:Q5_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf AGmind/agmind-rag-splitter-ru:Q5_K_M
# Run inference directly in the terminal:
./llama-cli -hf AGmind/agmind-rag-splitter-ru:Q5_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf AGmind/agmind-rag-splitter-ru:Q5_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf AGmind/agmind-rag-splitter-ru:Q5_K_M

Use Docker

docker model run hf.co/AGmind/agmind-rag-splitter-ru:Q5_K_M

LM Studio
Jan

vLLM

How to use AGmind/agmind-rag-splitter-ru with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "AGmind/agmind-rag-splitter-ru"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AGmind/agmind-rag-splitter-ru",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/AGmind/agmind-rag-splitter-ru:Q5_K_M

Ollama
How to use AGmind/agmind-rag-splitter-ru with Ollama:
```
ollama run hf.co/AGmind/agmind-rag-splitter-ru:Q5_K_M
```

Unsloth Studio

How to use AGmind/agmind-rag-splitter-ru with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for AGmind/agmind-rag-splitter-ru to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for AGmind/agmind-rag-splitter-ru to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for AGmind/agmind-rag-splitter-ru to start chatting

How to use AGmind/agmind-rag-splitter-ru with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf AGmind/agmind-rag-splitter-ru:Q5_K_M

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "AGmind/agmind-rag-splitter-ru:Q5_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use AGmind/agmind-rag-splitter-ru with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf AGmind/agmind-rag-splitter-ru:Q5_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default AGmind/agmind-rag-splitter-ru:Q5_K_M

Run Hermes

hermes

Atomic Chat new
Docker Model Runner
How to use AGmind/agmind-rag-splitter-ru with Docker Model Runner:
```
docker model run hf.co/AGmind/agmind-rag-splitter-ru:Q5_K_M
```

Lemonade

How to use AGmind/agmind-rag-splitter-ru with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull AGmind/agmind-rag-splitter-ru:Q5_K_M

Run and chat with the model

lemonade run user.agmind-rag-splitter-ru-Q5_K_M

List all available models

lemonade list

RU Context-Aware Document Splitter (T-lite-it-2.1 LoRA)

Дообучение t-tech/T-lite-it-2.1 (Qwen3-8B), которое режет русские документы на самодостаточные смысловые чанки для RAG, держа таблицы/код целыми. На вход — текст, заранее разбитый на нумерованные юниты; на выход — индексы границ + topic в JSON.

Использование

Это completion-модель, обученная на raw-Alpaca промпте (без чат-шаблона). Сначала разбей документ на нумерованные юниты; чанки собери на хосте по возвращённым индексам.

Промпт:

### Instruction:
Раздели документ на смысловые части для системы поиска (RAG). Каждая часть читается независимо, не разрывая предложений, таблиц и кода. Верни ТОЛЬКО номера единиц, после которых проходит граница, в формате JSON.

### Input:
[1] Первое предложение. [2] Второе. [3] | таблица |...|
### Response:

Вывод: {"splits": [2], "topic": "..."} — splits = индексы юнитов, после которых граница (1-индексные). Режь оригинал по этим точкам.

Полный пре-/пост-процессинг + рецепт сервинга на llama.cpp — в GitHub-репозитории.

Результаты (1500 holdout, согласие с метками учителя)

Валидный JSON	F1@0	F1@±1	exact-set
100%	0.656	0.821	29%

GGUF Q5_K_M совпадает с FP16 в пределах шума квантизации; работает на AMD Vulkan через llama.cpp.

Данные

Датасет: AGmind/agmind-rag-splitter-ru-data (~17k train + 12k синтетика, дистилляция от DeepSeek-V4-Flash).

Обучение

bf16 LoRA (r32, rsLoRA, all-linear, response-only) на RTX 5090; ~17k примеров дистилляции (DeepSeek-V4-Flash). См. docs/methodology.md в репо.

Файлы

LoRA-адаптер / merged FP16-веса
*-Q5_K_M.gguf (llama.cpp, Vulkan/CPU)

Ограничения

Метрики — согласие с учителем, не human-ground-truth. Лёгкая пере-сегментация. Для очень больших таблиц нужна отдельная табличная стратегия, а не эта боундари-модель.

Лицензия

Apache-2.0 (наследует лицензию базы T-lite-it-2.1).

Downloads last month: 17

Safetensors

Model size

8B params

Tensor type

BF16

Model tree for AGmind/agmind-rag-splitter-ru

Base model

Qwen/Qwen3-8B-Base

Finetuned

Qwen/Qwen3-8B

Finetuned

t-tech/T-lite-it-2.1

Adapter

(1)

this model

AGmind
/

agmind-rag-splitter-ru