# /// script # requires-python = ">=3.10,<3.14" # dependencies = [ # "accelerate", # "bitsandbytes>=0.43.0", # "datasets==4.3.0", # "hf_transfer", # "huggingface_hub>=0.34.0", # "marimo", # "peft", # "protobuf", # "sentencepiece", # "torchao>=0.16.0", # "transformers==4.56.2", # "triton>=3.2.0", # "trl==0.22.2", # "unsloth @ git+https://github.com/unslothai/unsloth", # "unsloth_zoo @ git+https://github.com/unslothai/unsloth-zoo", # ] # # [tool.uv] # no-build-package = [ # "bitsandbytes", # "triton", # "vllm", # "xformers", # ] # /// import marimo __generated_with = "0.23.8" app = marimo.App() @app.cell def _(): import marimo as mo return (mo,) @app.cell(hide_code=True) def _(mo): mo.md(r""" To run this notebook, hit the **▶ Run all** button in the bottom-right corner - or use `Ctrl/Cmd + Shift + R`.

Join Discord if you need help + ⭐ Star us on Github ⭐

To install Unsloth on your local device, follow [our guide](https://unsloth.ai/docs/get-started/install). This notebook is licensed [LGPL-3.0](https://github.com/unslothai/notebooks?tab=LGPL-3.0-1-ov-file#readme). You will learn how to do [data prep](#Data), how to [train](#Train), how to [run the model](#Inference), & how to save it """) return @app.cell(hide_code=True) def _(mo): mo.md(r""" ### News """) return @app.cell(hide_code=True) def _(mo): mo.md(r""" Introducing **Unsloth Studio** - a new open source, no-code web UI to train and run LLMs. [Blog](https://unsloth.ai/docs/new/studio) • [Notebook](https://github.com/unslothai/unsloth/blob/main/studio/Unsloth_Studio_Colab.ipynb)

_{Train models — no code needed}

_{Run GGUF models on Mac, Windows & Linux}

Train MoEs - DeepSeek, GLM, Qwen and gpt-oss 12x faster with 35% less VRAM. [Blog](https://unsloth.ai/docs/new/faster-moe) Ultra Long-Context Reinforcement Learning is here with 7x more context windows! [Blog](https://unsloth.ai/docs/new/grpo-long-context) New in Reinforcement Learning: [FP8 RL](https://unsloth.ai/docs/new/fp8-reinforcement-learning) • [Vision RL](https://unsloth.ai/docs/new/vision-reinforcement-learning-vlm-rl) • [Standby](https://unsloth.ai/docs/basics/memory-efficient-rl) • [gpt-oss RL](https://unsloth.ai/docs/new/gpt-oss-reinforcement-learning) Visit our docs for all our [model uploads](https://unsloth.ai/docs/get-started/unsloth-model-catalog) and [notebooks](https://unsloth.ai/docs/get-started/unsloth-notebooks). """) return @app.cell(hide_code=True) def _(mo): mo.md(r""" ### Unsloth """) return @app.cell def _(): from unsloth import FastLanguageModel import torch max_seq_length = 2048 # Choose any! We auto support RoPE Scaling internally! dtype = None # None for auto detection. Float16 for Tesla T4, V100, Bfloat16 for Ampere+ load_in_4bit = True # Use 4bit quantization to reduce memory usage. Can be False. # 4bit pre quantized models we support for 4x faster downloading + no OOMs. fourbit_models = [ "unsloth/mistral-7b-bnb-4bit", "unsloth/mistral-7b-instruct-v0.2-bnb-4bit", "unsloth/llama-2-7b-bnb-4bit", "unsloth/llama-2-13b-bnb-4bit", "unsloth/codellama-34b-bnb-4bit", "unsloth/tinyllama-bnb-4bit", "unsloth/gemma-7b-bnb-4bit", # New Google 6 trillion tokens model 2.5x faster! "unsloth/gemma-2b-bnb-4bit", ] # More models at https://huggingface.co/unsloth model, tokenizer = FastLanguageModel.from_pretrained( model_name="unsloth/mistral-7b-instruct-v0.3-bnb-4bit", # Choose ANY! eg teknium/OpenHermes-2.5-Mistral-7B max_seq_length=max_seq_length, # Choose any! We auto support RoPE Scaling internally! dtype=dtype, # None for auto detection. Float16 for Tesla T4, V100, Bfloat16 for Ampere+ load_in_4bit=load_in_4bit, # Use 4bit quantization to reduce memory usage. Can be False. # token = "YOUR_HF_TOKEN", # HF Token for gated models ) return ( FastLanguageModel, dtype, load_in_4bit, max_seq_length, model, tokenizer, torch, ) @app.cell(hide_code=True) def _(mo): mo.md(r""" We now add LoRA adapters so we only need to update 1 to 10% of all parameters! """) return @app.cell def _(FastLanguageModel, model): model_1 = FastLanguageModel.get_peft_model( model, r=16, # Choose any number > 0 ! Suggested 8, 16, 32, 64, 128 target_modules=[ "q_proj", "k_proj", "v_proj", "o_proj", "gate_proj", "up_proj", "down_proj", ], lora_alpha=16, lora_dropout=0, # Supports any, but = 0 is optimized bias="none", # Supports any, but = "none" is optimized use_gradient_checkpointing="unsloth", # True or "unsloth" for very long context random_state=3407, use_rslora=False, # We support rank stabilized LoRA loftq_config=None, # And LoftQ ) return (model_1,) @app.cell(hide_code=True) def _(mo): mo.md(r""" ### Data Prep We now use the `ChatML` format for conversation style finetunes. We use [Open Assistant conversations](https://huggingface.co/datasets/philschmid/guanaco-sharegpt-style) in ShareGPT style. ChatML renders multi turn conversations like below: ``` <|im_start|>system You are a helpful assistant.<|im_end|> <|im_start|>user What's the capital of France?<|im_end|> <|im_start|>assistant Paris. ``` **[NOTE]** To train only on completions (ignoring the user's input) read our docs [here](https://unsloth.ai/docs/get-started/fine-tuning-llms-guide/lora-hyperparameters-guide#training-on-completions-only-masking-out-inputs) We use our `get_chat_template` function to get the correct chat template. We support `zephyr, chatml, mistral, llama, alpaca, vicuna, vicuna_old` and our own optimized `unsloth` template. Normally one has to train `<|im_start|>` and `<|im_end|>`. We instead map `<|im_end|>` to be the EOS token, and leave `<|im_start|>` as is. This requires no additional training of additional tokens. Note ShareGPT uses `{"from": "human", "value" : "Hi"}` and not `{"role": "user", "content" : "Hi"}`, so we use `mapping` to map it. For text completions like novel writing, try this [notebook](https://github.com/unslothai/notebooks/blob/main/nb/Mistral_(7B)-Text_Completion.ipynb). """) return @app.cell def _(tokenizer): from unsloth.chat_templates import get_chat_template tokenizer_1 = get_chat_template( tokenizer, chat_template="chatml", # Supports zephyr, chatml, mistral, llama, alpaca, vicuna, vicuna_old, unsloth mapping={ # ShareGPT style "role": "from", "content": "value", "user": "human", "assistant": "gpt", }, map_eos_token=True, # Maps <|im_end|> to instead ) def formatting_prompts_func(examples): convos = examples[ "conversations" ] # Supports zephyr, chatml, mistral, llama, alpaca, vicuna, vicuna_old, unsloth texts = [ tokenizer_1.apply_chat_template( convo, tokenize=False, add_generation_prompt=False ) for convo in convos ] # ShareGPT style return {"text": texts} # Maps <|im_end|> to instead from datasets import load_dataset dataset = load_dataset("philschmid/guanaco-sharegpt-style", split="train") dataset = dataset.map(formatting_prompts_func, batched=True) return dataset, get_chat_template, tokenizer_1 @app.cell(hide_code=True) def _(mo): mo.md(r""" Let's see how the `ChatML` format works by printing the 5th element """) return @app.cell def _(dataset): dataset[5]["conversations"] return @app.cell def _(dataset): print(dataset[5]["text"]) return @app.cell(hide_code=True) def _(mo): mo.md(r""" If you're looking to make your own chat template, that also is possible! You must use the Jinja templating regime. We provide our own stripped down version of the `Unsloth template` which we find to be more efficient, and leverages ChatML, Zephyr and Alpaca styles. More info on chat templates on [our wiki page!](https://github.com/unslothai/unsloth/wiki#chat-templates) """) return @app.cell def _(get_chat_template): unsloth_template = ( "{{ bos_token }}" "{{ 'You are a helpful assistant to the user\n' }}" "{% for message in messages %}" "{% if message['role'] == 'user' %}" "{{ '>>> User: ' + message['content'] + '\n' }}" "{% elif message['role'] == 'assistant' %}" "{{ '>>> Assistant: ' + message['content'] + eos_token + '\n' }}" "{% endif %}" "{% endfor %}" "{% if add_generation_prompt %}" "{{ '>>> Assistant: ' }}" "{% endif %}" ) unsloth_eos_token = "eos_token" if False: _tokenizer = get_chat_template( _tokenizer, chat_template=(unsloth_template, unsloth_eos_token), # Supports zephyr, chatml, mistral, llama, alpaca, vicuna, vicuna_old, unsloth mapping={ # ShareGPT style "role": "from", "content": "value", "user": "human", "assistant": "gpt", }, map_eos_token=True, # Maps <|im_end|> to instead ) return @app.cell(hide_code=True) def _(mo): mo.md(r""" ### Train the model Now let's train our model. We do 60 steps to speed things up, but you can set `num_train_epochs=1` for a full run, and turn off `max_steps=None`. We also support `DPOTrainer` and `GRPOTrainer` for reinforcement learning!! """) return @app.cell def _(dataset, max_seq_length, model_1, tokenizer_1): from trl import SFTConfig, SFTTrainer trainer = SFTTrainer( model=model_1, tokenizer=tokenizer_1, train_dataset=dataset, dataset_text_field="text", max_seq_length=max_seq_length, # Choose any! We auto support RoPE Scaling internally! packing=False, # Can make training 5x faster for short sequences. args=SFTConfig( per_device_train_batch_size=2, gradient_accumulation_steps=4, warmup_steps=5, max_steps=60, learning_rate=0.0002, logging_steps=1, optim="adamw_8bit", weight_decay=0.001, lr_scheduler_type="linear", seed=3407, output_dir="outputs", report_to="none", # Use TrackIO/WandB etc ), ) return (trainer,) @app.cell def _(torch): # @title Show current memory stats gpu_stats = torch.cuda.get_device_properties(0) start_gpu_memory = round(torch.cuda.max_memory_reserved() / 1024 / 1024 / 1024, 3) max_memory = round(gpu_stats.total_memory / 1024 / 1024 / 1024, 3) print(f"GPU = {gpu_stats.name}. Max memory = {max_memory} GB.") print(f"{start_gpu_memory} GB of memory reserved.") return max_memory, start_gpu_memory @app.cell def _(trainer): trainer_stats = trainer.train() return (trainer_stats,) @app.cell def _(max_memory, start_gpu_memory, torch, trainer_stats): # @title Show final memory and time stats used_memory = round(torch.cuda.max_memory_reserved() / 1024 / 1024 / 1024, 3) used_memory_for_lora = round(used_memory - start_gpu_memory, 3) used_percentage = round(used_memory / max_memory * 100, 3) lora_percentage = round(used_memory_for_lora / max_memory * 100, 3) print(f"{trainer_stats.metrics['train_runtime']} seconds used for training.") print( f"{round(trainer_stats.metrics['train_runtime'] / 60, 2)} minutes used for training." ) print(f"Peak reserved memory = {used_memory} GB.") print(f"Peak reserved memory for training = {used_memory_for_lora} GB.") print(f"Peak reserved memory % of max memory = {used_percentage} %.") print(f"Peak reserved memory for training % of max memory = {lora_percentage} %.") return @app.cell(hide_code=True) def _(mo): mo.md(r""" ### Inference Let's run the model! Since we're using `ChatML`, use `apply_chat_template` with `add_generation_prompt` set to `True` for inference. """) return @app.cell def _(FastLanguageModel, get_chat_template, model_1, tokenizer_1): tokenizer_2 = get_chat_template( tokenizer_1, chat_template="chatml", # Supports zephyr, chatml, mistral, llama, alpaca, vicuna, vicuna_old, unsloth mapping={ # ShareGPT style "role": "from", "content": "value", "user": "human", "assistant": "gpt", }, map_eos_token=True, # Maps <|im_end|> to instead ) FastLanguageModel.for_inference(model_1) messages = [ {"from": "human", "value": "Continue the fibonacci sequence: 1, 1, 2, 3, 5, 8,"} ] inputs = tokenizer_2.apply_chat_template( messages, tokenize=True, add_generation_prompt=True, return_tensors="pt" ).to("cuda") outputs = model_1.generate( input_ids=inputs, max_new_tokens=64, use_cache=True ) # Supports zephyr, chatml, mistral, llama, alpaca, vicuna, vicuna_old, unsloth tokenizer_2.batch_decode( outputs ) return (tokenizer_2,) @app.cell(hide_code=True) def _(mo): mo.md(r""" You can also use a `TextStreamer` for continuous inference - so you can see the generation token by token, instead of waiting the whole time! """) return @app.cell def _(FastLanguageModel, model_1, tokenizer_2): FastLanguageModel.for_inference(model_1) # Enable native 2x faster inference messages_1 = [ {"from": "human", "value": "Continue the fibonacci sequence: 1, 1, 2, 3, 5, 8,"} ] inputs_1 = tokenizer_2.apply_chat_template( messages_1, tokenize=True, add_generation_prompt=True, return_tensors="pt" ).to("cuda") from transformers import TextStreamer text_streamer = TextStreamer(tokenizer_2) _ = model_1.generate( input_ids=inputs_1, streamer=text_streamer, max_new_tokens=128, use_cache=True ) # Must add for generation return (TextStreamer,) @app.cell(hide_code=True) def _(mo): mo.md(r""" ### Saving, loading finetuned models To save the final model as LoRA adapters, either use Hugging Face's `push_to_hub` for an online save or `save_pretrained` for a local save. **[NOTE]** This ONLY saves the LoRA adapters, and not the full model. To save to 16bit or GGUF, scroll down! """) return @app.cell def _(model_1, tokenizer_2): model_1.save_pretrained("mistral_v0_lora") tokenizer_2.save_pretrained("mistral_v0_lora") return @app.cell(hide_code=True) def _(mo): mo.md(r""" Now if you want to load the LoRA adapters we just saved for inference, set `False` to `True`: """) return @app.cell def _(TextStreamer, dtype, load_in_4bit, max_seq_length, model_1, tokenizer_2): if False: from unsloth import FastLanguageModel as _FastLanguageModel _model, _tokenizer = _FastLanguageModel.from_pretrained( model_name="mistral_v0_lora", # Choose ANY! eg teknium/OpenHermes-2.5-Mistral-7B max_seq_length=max_seq_length, # Choose any! We auto support RoPE Scaling internally! dtype=dtype, # None for auto detection. Float16 for Tesla T4, V100, Bfloat16 for Ampere+ load_in_4bit=load_in_4bit, # Use 4bit quantization to reduce memory usage. Can be False. ) _FastLanguageModel.for_inference(_model) messages_2 = [{"from": "human", "value": "What is a famous tall tower in Paris?"}] inputs_2 = tokenizer_2.apply_chat_template( messages_2, tokenize=True, add_generation_prompt=True, return_tensors="pt" ).to("cuda") text_streamer_1 = TextStreamer(tokenizer_2) _ = model_1.generate( input_ids=inputs_2, streamer=text_streamer_1, max_new_tokens=128, use_cache=True ) # Must add for generation return @app.cell(hide_code=True) def _(mo): mo.md(r""" You can also use Hugging Face's `AutoPeftModelForCausalLM`. Only use this if you do not have `unsloth` installed. It can be hopelessly slow, since `4bit` model downloading is not supported, and Unsloth's **inference is 2x faster**. """) return @app.cell def _(load_in_4bit): if False: from peft import AutoPeftModelForCausalLM as _AutoPeftModelForCausalLM from transformers import AutoTokenizer as _AutoTokenizer _model = _AutoPeftModelForCausalLM.from_pretrained( "mistral_v0_lora", load_in_4bit=load_in_4bit ) _tokenizer = _AutoTokenizer.from_pretrained("mistral_v0_lora") return @app.cell(hide_code=True) def _(mo): mo.md(r""" ### Saving to float16 for VLLM We also support saving to `float16` directly. Select `merged_16bit` for float16 or `merged_4bit` for int4. We also allow `lora` adapters as a fallback. Use `push_to_hub_merged` to upload to your Hugging Face account! You can go to https://huggingface.co/settings/tokens for your personal tokens. See [our docs](https://unsloth.ai/docs/basics/inference-and-deployment) for more deployment options. """) return @app.cell def _(model_1, tokenizer_2): # Merge to 16bit if False: model_1.save_pretrained_merged( "mistral_v0_finetune_16bit", tokenizer_2, save_method="merged_16bit" ) if False: # Merge to 4bit model_1.push_to_hub_merged( "HF_USERNAME/mistral_v0_finetune_16bit", tokenizer_2, save_method="merged_16bit", token="YOUR_HF_TOKEN", ) if False: model_1.save_pretrained_merged( "mistral_v0_finetune_4bit", tokenizer_2, save_method="merged_4bit" ) # Just LoRA adapters if False: model_1.push_to_hub_merged( "HF_USERNAME/mistral_v0_finetune_4bit", tokenizer_2, save_method="merged_4bit", token="YOUR_HF_TOKEN", ) if False: model_1.save_pretrained("mistral_v0_lora") tokenizer_2.save_pretrained("mistral_v0_lora") if False: model_1.push_to_hub("HF_USERNAME/mistral_v0_lora", token="YOUR_HF_TOKEN") tokenizer_2.push_to_hub("HF_USERNAME/mistral_v0_lora", token="YOUR_HF_TOKEN") return @app.cell(hide_code=True) def _(mo): mo.md(r""" ### GGUF / llama.cpp Conversion To save to `GGUF` / `llama.cpp`, we support it natively now! We clone `llama.cpp` and we default save it to `q8_0`. We allow all methods like `q4_k_m`. Use `save_pretrained_gguf` for local saving and `push_to_hub_gguf` for uploading to HF. Some supported quant methods (full list on our [docs page](https://unsloth.ai/docs/basics/inference-and-deployment/saving-to-gguf)): * `q8_0` - Fast conversion. High resource use, but generally acceptable. * `q4_k_m` - Recommended. Uses Q6_K for half of the attention.wv and feed_forward.w2 tensors, else Q4_K. * `q5_k_m` - Recommended. Uses Q6_K for half of the attention.wv and feed_forward.w2 tensors, else Q5_K. """) return @app.cell def _(model_1, tokenizer_2): # Save to 8bit Q8_0 if False: model_1.save_pretrained_gguf("mistral_v0_finetune", tokenizer_2) if False: # Save to 16bit GGUF model_1.push_to_hub_gguf( "HF_USERNAME/mistral_v0_finetune", tokenizer_2, token="YOUR_HF_TOKEN" ) if False: model_1.save_pretrained_gguf( "mistral_v0_finetune", tokenizer_2, quantization_method="f16" ) # Save to q4_k_m GGUF if False: model_1.push_to_hub_gguf( "HF_USERNAME/mistral_v0_finetune", tokenizer_2, quantization_method="f16", token="YOUR_HF_TOKEN", ) if False: model_1.save_pretrained_gguf( "mistral_v0_finetune", tokenizer_2, quantization_method="q4_k_m" ) if False: model_1.push_to_hub_gguf( "HF_USERNAME/mistral_v0_finetune", tokenizer_2, quantization_method="q4_k_m", token="YOUR_HF_TOKEN", ) return @app.cell(hide_code=True) def _(mo): mo.md(r""" And we're done! If you have any questions on Unsloth, we have a [Discord](https://discord.gg/unsloth) channel! If you find any bugs or want to keep updated with the latest LLM stuff, or need help, join projects etc, feel free to join our Discord! Some other resources: 1. Looking to use Unsloth locally? Read our [Installation Guide](https://unsloth.ai/docs/get-started/install) for details on installing Unsloth on Windows, Docker, AMD, Intel GPUs. 2. Learn how to do Reinforcement Learning with our [RL Guide and notebooks](https://unsloth.ai/docs/get-started/reinforcement-learning-rl-guide). 3. Read our guides and notebooks for [Text-to-speech (TTS)](https://unsloth.ai/docs/basics/text-to-speech-tts-fine-tuning) and [vision](https://unsloth.ai/docs/basics/vision-fine-tuning) model support. 4. Explore our [LLM Tutorials Directory](https://unsloth.ai/docs/models/tutorials-how-to-fine-tune-and-run-llms) to find dedicated guides for each model. 5. Need help with Inference? Read our [Inference & Deployment page](https://unsloth.ai/docs/basics/inference-and-deployment) for details on using vLLM, llama.cpp, Ollama etc.

Join Discord if you need help + ⭐️ Star us on Github ⭐️ This notebook and all Unsloth notebooks are licensed [LGPL-3.0](https://github.com/unslothai/notebooks?tab=LGPL-3.0-1-ov-file#readme)

""") return if __name__ == "__main__": app.run()