Instructions to use Rohanify/Indenta-13M-Python with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Rohanify/Indenta-13M-Python with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="Rohanify/Indenta-13M-Python",
	filename="indenta-9m.gguf",
)

output = llm(
	"Once upon a time,",
	max_tokens=512,
	echo=True
)
print(output)

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use Rohanify/Indenta-13M-Python with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf Rohanify/Indenta-13M-Python
# Run inference directly in the terminal:
llama-cli -hf Rohanify/Indenta-13M-Python

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf Rohanify/Indenta-13M-Python
# Run inference directly in the terminal:
llama-cli -hf Rohanify/Indenta-13M-Python

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 Rohanify/Indenta-13M-Python
# Run inference directly in the terminal:
./llama-cli -hf Rohanify/Indenta-13M-Python

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 Rohanify/Indenta-13M-Python
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Rohanify/Indenta-13M-Python

Use Docker

docker model run hf.co/Rohanify/Indenta-13M-Python

LM Studio
Jan

vLLM

How to use Rohanify/Indenta-13M-Python with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Rohanify/Indenta-13M-Python"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Rohanify/Indenta-13M-Python",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Rohanify/Indenta-13M-Python

Ollama
How to use Rohanify/Indenta-13M-Python with Ollama:
```
ollama run hf.co/Rohanify/Indenta-13M-Python
```

Unsloth Studio new

How to use Rohanify/Indenta-13M-Python 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 Rohanify/Indenta-13M-Python 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 Rohanify/Indenta-13M-Python to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for Rohanify/Indenta-13M-Python to start chatting

Docker Model Runner
How to use Rohanify/Indenta-13M-Python with Docker Model Runner:
```
docker model run hf.co/Rohanify/Indenta-13M-Python
```

Lemonade

How to use Rohanify/Indenta-13M-Python with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Rohanify/Indenta-13M-Python

Run and chat with the model

lemonade run user.Indenta-13M-Python-{{QUANT_TAG}}

List all available models

lemonade list

Rohanify commited on 4 days ago

Commit

050df37

verified ·

1 Parent(s): 34186d6

Update README.md

Browse files

Files changed (1) hide show

README.md +17 -32

README.md CHANGED Viewed

@@ -13,18 +13,24 @@ tags:
 pipeline_tag: text-generation
 ollama:
   template: |
-    ### Instruction:
     {{ .Prompt }}
     ### Response:
   params:
     temperature: 0.1
-    top_p: 0.8
     repeat_penalty: 2.0
-    repeat_last_n: 64
-    num_ctx: 256
     stop:
-    - "### Instruction:"
     - "### Response:"
     - "<|endoftext|>"
 ---
@@ -36,33 +42,12 @@ This model is built specifically for lightning-fast Python code completions and
 ---
-## 🛠️ How to Run Instantly via Ollama
-Because the system configuration is baked directly into this Hugging Face repository card, nobody needs to manually create a local `Modelfile`.
-### ⚠️ Critical Usage Note for 13M Parameters
-Because this model is highly optimized and ultra-lightweight (~13M parameters), it is architectural design-limited to **single-turn tasks**. It does not possess a multi-turn chat memory tracking mechanism.
-Allowing Ollama to stack message history will cause the attention layers to collapse.
-To ensure a perfect experience without crashes, use either of the two methods below:
-### Method 1: Stateless Mode (Recommended 🚀)
-Pass your prompt directly inside quotes in your terminal. This forces Ollama to run a clean, stateless, single-turn generation that will **never crash**:
-```bash
-ollama run hf.co/Rohanify/Indenta-13M-Python "write a function to reverse a list"
-```
-### Method 2: Interactive Terminal Mode
-If you are using the interactive chat loop (`ollama run hf.co/Rohanify/Indenta-13M-Python`), simply wipe the conversation memory before typing your next prompt by entering `/clear`:
-```text
->>> write a for loop
-[Model generates code safely]
->>> /clear
-Cleared session context
->>> reverse a linked list
-[Model generates next code safely]
-```

 pipeline_tag: text-generation
 ollama:
   template: |
+    {{- if .Prompt }}## Instruction:
     {{ .Prompt }}
+    ### Input:
+    Processing your request...
+    {{ end -}}
     ### Response:
+    {{- if .Response }}{{ .Response }}{{ end -}}
   params:
     temperature: 0.1
+    top_p: 0.85
     repeat_penalty: 2.0
+    repeat_last_n: 32
+    num_ctx: 512
     stop:
+    - "## Instruction:"
+    - "### Input:"
     - "### Response:"
     - "<|endoftext|>"
 ---
 ---
+## 🛠️ Web UI & Multi-Turn Stability Note
+This model features a hyper-lightweight 13M parameter footprint optimized for single-task completions based directly on structural templates.
+Because it lacks the large capacity required for conversational context processing, it can drop formatting structure if a graphical Web UI forces a massive, multi-message chat stream into its context layers.
+### For Best Performance in Web UIs:
+1. **Use New Chat Threads:** Click the "New Chat" or "Clear" button in your user interface between coding tasks. This keeps the model's focus squarely on your active prompt.
+2. **Synchronized Template:** This model card includes an aligned layout format (`## Instruction:` and `### Input:`) matching the training data to stop token bleeding across chat iterations.