Instructions to use pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF 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 pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_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 pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_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 pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_K_M

Use Docker

docker model run hf.co/pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_K_M

Ollama
How to use pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF with Ollama:
```
ollama run hf.co/pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_K_M
```
Unsloth Desktop

How to use pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF with Docker Model Runner:
```
docker model run hf.co/pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_K_M
```

Lemonade

How to use pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.qwen3.5-9b-hindi-instruct-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_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 pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

pankajpandey-dev commited on Jul 4

Commit

0741477

verified ·

1 Parent(s): fdf65f6

Upload README.md with huggingface_hub

Browse files

Files changed (1) hide show

README.md +25 -49

README.md CHANGED Viewed

@@ -1,66 +1,42 @@
 ---
 license: apache-2.0
-language:
-- hi
-- en
 base_model: pankajpandey-dev/qwen3.5-9b-hindi-instruct
 pipeline_tag: text-generation
-tags:
-- hindi
-- qwen3_5
-- qwen3.5
-- gguf
-- llama.cpp
-- lmstudio
-- unsloth
-- conversational
-- india
 ---
-# 🇮🇳 Qwen3.5-9B-Hindi-Instruct — GGUF
-CPU-friendly quants of [qwen3.5-9b-hindi-instruct](https://huggingface.co/pankajpandey-dev/qwen3.5-9b-hindi-instruct): a Qwen3.5-9B fine-tune that answers **instantly in pure Devanagari Hindi** — no English chain-of-thought, no code-switching. Full before/after, data mix, and recipe in the main card.
-## Files
-| File | Quant | Size | For |
 |---|---|---|---|
-| Qwen3.5-9B-Hindi-Instruct-Q4_K_M.gguf | Q4_K_M | 5.4 GB | **Default.** CPU / 8GB+ RAM |
-| Qwen3.5-9B-Hindi-Instruct-Q5_K_M.gguf | Q5_K_M | 6.2 GB | Better quality, still CPU-viable |
-| Qwen3.5-9B-Hindi-Instruct-Q8_0.gguf | Q8_0 | 9.2 GB | Near-lossless |
-Text-only — no mmproj (vision layers not fine-tuned or tested).
 ## Run it
-**llama.cpp**
-```bash
-llama-cli -m Qwen3.5-9B-Hindi-Instruct-Q4_K_M.gguf --jinja -p "जल संरक्षण के तीन तरीके बताइए।" -n 300
 ```
-**LM Studio:** search this repo, download Q4_K_M, chat. Correct template is embedded.
-**Ollama (pull from HF):**
-```bash
-ollama run hf.co/pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_K_M
 ```
-Requires recent Ollama with qwen3.5 arch support; "ollama create" from local GGUFs may not work — pull from HF instead.
-**Python (llama-cpp-python >= 0.3.32 — older versions lack qwen3.5 arch support):**
 ```python
 from llama_cpp import Llama
-llm = Llama.from_pretrained(repo_id="pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF",
-                            filename="*Q4_K_M*", n_ctx=4096)
-out = llm.create_chat_completion(messages=[{"role": "user", "content": "चाय बनाने की विधि बताइए।"}])
-print(out["choices"][0]["message"]["content"])
 ```
-## ⚠️ Template note
-The embedded template **always emits a closed empty <think> block** — direct answers in every runtime. Deliberate: llama.cpp sets enable_thinking=true by default, which on the stock template re-triggers English chain-of-thought. Want thinking? Use base Qwen3.5-9B.
-## Limitations
-Native Hindi instruction-follower, not an encyclopedia — long-form factual content can contain errors. Q4 trades a little quality for size. Full list in the main card.
-Part of the [🇮🇳 Hindi LLM Series](https://huggingface.co/pankajpandey-dev).

 ---
 license: apache-2.0
+language: [hi, en]
 base_model: pankajpandey-dev/qwen3.5-9b-hindi-instruct
+base_model_relation: quantized
 pipeline_tag: text-generation
+tags: [hindi, qwen3.5, gguf, llama.cpp, lmstudio, quantized, india]
 ---
+# Qwen3.5-9B Hindi Instruct — GGUF 🇮🇳
+**Native-Hindi Qwen3.5-9B for your laptop.** Quantized from [pankajpandey-dev/qwen3.5-9b-hindi-instruct](https://huggingface.co/pankajpandey-dev/qwen3.5-9b-hindi-instruct). Answers directly in Devanagari — no English thinking, no code-switching.
+## Which file?
+| File | Size | RAM needed | Use when |
 |---|---|---|---|
+| Q4_K_M | ~5.7 GB | ~8 GB | **Recommended** — best size/quality on CPU |
+| Q5_K_M | ~6.7 GB | ~9 GB | More quality headroom |
+| Q8_0 | ~10 GB | ~12 GB | Near-lossless |
 ## Run it
+**llama.cpp** (build from ~March 2026 or newer for Qwen3.5 support):
 ```
+llama-cli -m qwen3.5-9b-hindi-Q4_K_M.gguf --jinja -cnv --repeat-penalty 1.1
 ```
+**LM Studio:** search this repo name, download Q4_K_M, set repeat penalty 1.1, chat.
+**Python (llama-cpp-python 0.3.32 or newer):**
 ```python
 from llama_cpp import Llama
+llm = Llama(model_path="qwen3.5-9b-hindi-Q4_K_M.gguf", n_ctx=4096)
+r = llm.create_chat_completion(
+    messages=[{"role": "user", "content": "योग के चार लाभ बताइए।"}],
+    repeat_penalty=1.1)
+print(r["choices"][0]["message"]["content"])
 ```
+**Ollama:** `ollama run hf.co/pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF:Q4_K_M` — needs an Ollama build with Qwen3.5 support; if it errors, use llama.cpp or LM Studio.
+Tip: use repeat penalty 1.1 — long letter-style outputs can loop without it.
+---
+## 🇮🇳 About the Hindi LLM Series
+Weekly open releases making small LLMs speak fluent, native Hindi — trained on free/low-cost GPUs, shipped as GGUF for laptops and edge devices. Built by [pankajpandey-dev](https://huggingface.co/pankajpandey-dev) *(contact links on profile)*.
+**This release:** [Model](https://huggingface.co/pankajpandey-dev/qwen3.5-9b-hindi-instruct) · [GGUF](https://huggingface.co/pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF) · [LoRA](https://huggingface.co/pankajpandey-dev/qwen3.5-9b-hindi-instruct-lora) · **Series:** [🇮🇳 Hindi LLM Collection](https://huggingface.co/pankajpandey-dev)