Instructions to use comarproject/lale-9b-2603 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use comarproject/lale-9b-2603 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="comarproject/lale-9b-2603")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("comarproject/lale-9b-2603", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use comarproject/lale-9b-2603 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 comarproject/lale-9b-2603:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf comarproject/lale-9b-2603:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf comarproject/lale-9b-2603:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf comarproject/lale-9b-2603: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 comarproject/lale-9b-2603:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf comarproject/lale-9b-2603: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 comarproject/lale-9b-2603:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf comarproject/lale-9b-2603:Q4_K_M

Use Docker

docker model run hf.co/comarproject/lale-9b-2603:Q4_K_M

LM Studio
Jan

vLLM

How to use comarproject/lale-9b-2603 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "comarproject/lale-9b-2603"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "comarproject/lale-9b-2603",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/comarproject/lale-9b-2603:Q4_K_M

SGLang

How to use comarproject/lale-9b-2603 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "comarproject/lale-9b-2603" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "comarproject/lale-9b-2603",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "comarproject/lale-9b-2603" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "comarproject/lale-9b-2603",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use comarproject/lale-9b-2603 with Ollama:
```
ollama run hf.co/comarproject/lale-9b-2603:Q4_K_M
```

Unsloth Studio

How to use comarproject/lale-9b-2603 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 comarproject/lale-9b-2603 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 comarproject/lale-9b-2603 to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for comarproject/lale-9b-2603 to start chatting

How to use comarproject/lale-9b-2603 with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf comarproject/lale-9b-2603:Q4_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": "comarproject/lale-9b-2603:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

OpenClaw new

How to use comarproject/lale-9b-2603 with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf comarproject/lale-9b-2603: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 "comarproject/lale-9b-2603: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"

Docker Model Runner
How to use comarproject/lale-9b-2603 with Docker Model Runner:
```
docker model run hf.co/comarproject/lale-9b-2603:Q4_K_M
```

Lemonade

How to use comarproject/lale-9b-2603 with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull comarproject/lale-9b-2603:Q4_K_M

Run and chat with the model

lemonade run user.lale-9b-2603-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use comarproject/lale-9b-2603 with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf comarproject/lale-9b-2603: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 comarproject/lale-9b-2603:Q4_K_M

Run Hermes

hermes

Atomic Chat

comarproject commited on Mar 18

Commit

ba72741

verified ·

1 Parent(s): eda4955

Add model card with training details and benchmarks

Browse files

Files changed (1) hide show

README.md +10 -10

README.md CHANGED Viewed

@@ -17,7 +17,7 @@ tags:
   - qwen3.5
 pipeline_tag: text-generation
 model-index:
-  - name: lale-9b-v2
     results:
       - task:
           type: text-generation
@@ -40,7 +40,7 @@ model-index:
             value: 0.376
 ---
-# lale-9b-v2
 **lale** (Turkish for "tulip") is a Turkish instruction-following language model fine-tuned from [Qwen3.5-9B](https://huggingface.co/Qwen/Qwen3.5-9B). It is designed to be the best Turkish language model at its size class, with strong performance in general knowledge, reasoning, tool use, grammar, finance, and legal domains.
@@ -81,9 +81,9 @@ All data was filtered for format validity, length bounds, exact deduplication, a
 Evaluated using the [terazi](https://github.com/selimozten/terazi) Turkish language model benchmark suite.
-### v1 vs v2 Comparison
-| Category | v1 (98K data) | v2 (118K data) | Change |
 |---|---|---|---|
 | **core** | 0.511 | **0.516** | +1.0% |
 | common_sense | 0.970 | **0.980** | +1.0% |
@@ -138,13 +138,13 @@ ollama run lale
 from transformers import AutoModelForCausalLM, AutoTokenizer
 model = AutoModelForCausalLM.from_pretrained(
-    "comarproject/lale-9b-v2",
     subfolder="merged",
     torch_dtype="bfloat16",
     device_map="auto",
 )
 tokenizer = AutoTokenizer.from_pretrained(
-    "comarproject/lale-9b-v2",
     subfolder="merged",
 )
@@ -169,7 +169,7 @@ print(tokenizer.decode(outputs[0], skip_special_tokens=True))
 - Trained primarily on synthetic data from Claude models; may reflect Claude's style and biases
 - Context window limited to 2048 tokens during training (base model supports 128K)
-- Sentiment analysis regressed from v1 (-20%) -- may need targeted data for this subcategory
 - Some long legal/financial prompts may exceed the trained context length
 ## License
@@ -179,10 +179,10 @@ Apache 2.0
 ## Citation
 ```bibtex
-@misc{lale-9b-v2,
-  title={lale-9b-v2: Turkish Instruction Model Distilled from Frontier Models},
   author={Selim Ozten},
   year={2026},
-  url={https://huggingface.co/comarproject/lale-9b-v2}
 }
 ```

   - qwen3.5
 pipeline_tag: text-generation
 model-index:
+  - name: lale-9b-2603
     results:
       - task:
           type: text-generation
             value: 0.376
 ---
+# lale-9b-2603
 **lale** (Turkish for "tulip") is a Turkish instruction-following language model fine-tuned from [Qwen3.5-9B](https://huggingface.co/Qwen/Qwen3.5-9B). It is designed to be the best Turkish language model at its size class, with strong performance in general knowledge, reasoning, tool use, grammar, finance, and legal domains.
 Evaluated using the [terazi](https://github.com/selimozten/terazi) Turkish language model benchmark suite.
+### lale-9b-2602 vs lale-9b-2603
+| Category | 2602 (98K data) | 2603 (118K data) | Change |
 |---|---|---|---|
 | **core** | 0.511 | **0.516** | +1.0% |
 | common_sense | 0.970 | **0.980** | +1.0% |
 from transformers import AutoModelForCausalLM, AutoTokenizer
 model = AutoModelForCausalLM.from_pretrained(
+    "comarproject/lale-9b-2603",
     subfolder="merged",
     torch_dtype="bfloat16",
     device_map="auto",
 )
 tokenizer = AutoTokenizer.from_pretrained(
+    "comarproject/lale-9b-2603",
     subfolder="merged",
 )
 - Trained primarily on synthetic data from Claude models; may reflect Claude's style and biases
 - Context window limited to 2048 tokens during training (base model supports 128K)
+- Sentiment analysis regressed from 2602 (-20%) -- may need targeted data for this subcategory
 - Some long legal/financial prompts may exceed the trained context length
 ## License
 ## Citation
 ```bibtex
+@misc{lale-9b-2603,
+  title={lale-9b-2603: Turkish Instruction Model Distilled from Frontier Models},
   author={Selim Ozten},
   year={2026},
+  url={https://huggingface.co/comarproject/lale-9b-2603}
 }
 ```