app.py

!pip install "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
!pip install --no-deps xformers trl peft accelerate bitsandbytes

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-v0.3-bnb-4bit",      # New Mistral v3 2x faster!
    "unsloth/mistral-7b-instruct-v0.3-bnb-4bit",
    "unsloth/llama-3-8b-bnb-4bit",           # Llama-3 15 trillion tokens model 2x faster!
    "unsloth/llama-3-8b-Instruct-bnb-4bit",
    "unsloth/llama-3-70b-bnb-4bit",
    "unsloth/Phi-3-mini-4k-instruct",        # Phi-3 2x faster!
    "unsloth/Phi-3-medium-4k-instruct",
    "unsloth/mistral-7b-bnb-4bit",
    "unsloth/gemma-7b-bnb-4bit",             # Gemma 2.2x faster!
] # More models at https://huggingface.co./unsloth

model, tokenizer = FastLanguageModel.from_pretrained(
    model_name = "unsloth/llama-3-8b-bnb-4bit",
    max_seq_length = max_seq_length,
    dtype = dtype,
    load_in_4bit = load_in_4bit,
    # token = "hf_...", # use one if using gated models like meta-llama/Llama-2-7b-hf
)

from google.colab import drive
drive.mount('/content/drive')

import pandas as pd
df = pd.read_csv('/content/drive/MyDrive/Colab Notebooks/qa_examples.csv')
df.head(5)

df.columns = df.columns.str.strip()
df.columns

# Format into new columns
df['instruction'] = df.apply(lambda row: f"The following question is solved for {row['marks_available']} marks: {row['question']}. Referring to the mark-scheme, award the appropriate amount of marks to the student: {row['mark_scheme']}", axis=1)
df['input'] = df['student_response']
df['output'] = df.apply(lambda row: str({'marks': row['marks_award'], 'explanation': row['explanation']}), axis=1)

# Create a new DataFrame with the desired structure
formatted_df = pd.DataFrame({
    'instruction': df['instruction'],
    'input': df['input'],
    'output': df['output']
})

# Display the formatted DataFrame
formatted_df.head(5)

"""* We support Llama, Mistral, Phi-3, Gemma, Yi, DeepSeek, Qwen, TinyLlama, Vicuna, Open Hermes etc
* We support 16bit LoRA or 4bit QLoRA. Both 2x faster.
* `max_seq_length` can be set to anything, since we do automatic RoPE Scaling via [kaiokendev's](https://kaiokendev.github.io/til) method.
* With [PR 26037](https://github.com/huggingface/transformers/pull/26037), we support downloading 4bit models **4x faster**! [Our repo](https://huggingface.co./unsloth) has Llama, Mistral 4bit models.
* [**NEW**] We make Phi-3 Medium / Mini **2x faster**! See our [Phi-3 Medium notebook](https://colab.research.google.com/drive/1hhdhBa1j_hsymiW9m-WzxQtgqTH_NHqi?usp=sharing)

We now add LoRA adapters so we only need to update 1 to 10% of all parameters!
"""

model = 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
    # [NEW] "unsloth" uses 30% less VRAM, fits 2x larger batch sizes!
    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
)

"""<a name="Data"></a>
### Data Prep
We now use the Alpaca dataset from [yahma](https://huggingface.co./datasets/yahma/alpaca-cleaned), which is a filtered version of 52K of the original [Alpaca dataset](https://crfm.stanford.edu/2023/03/13/alpaca.html). You can replace this code section with your own data prep.

**[NOTE]** To train only on completions (ignoring the user's input) read TRL's docs [here](https://huggingface.co./docs/trl/sft_trainer#train-on-completions-only).

**[NOTE]** Remember to add the **EOS_TOKEN** to the tokenized output!! Otherwise you'll get infinite generations!

If you want to use the `llama-3` template for ShareGPT datasets, try our conversational [notebook](https://colab.research.google.com/drive/1XamvWYinY6FOSX9GLvnqSjjsNflxdhNc?usp=sharing).

For text completions like novel writing, try this [notebook](https://colab.research.google.com/drive/1ef-tab5bhkvWmBOObepl1WgJvfvSzn5Q?usp=sharing).

<a name="Train"></a>
### Train the model
Now let's use Huggingface TRL's `SFTTrainer`! More docs here: [TRL SFT docs](https://huggingface.co./docs/trl/sft_trainer). 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 TRL's `DPOTrainer`!
"""

alpaca_prompt = """
### Instruction:
{}
### Input:
{}
### Response:
{}"""

EOS_TOKEN = tokenizer.eos_token # Must add EOS_TOKEN
def formatting_prompts_func(examples):
    instructions = examples["instruction"]
    inputs       = examples["input"]
    outputs      = examples["output"]
    texts = []
    for instruction, input, output in zip(instructions, inputs, outputs):
        text = alpaca_prompt.format(instruction, input, output) + EOS_TOKEN
        texts.append(text)
    return { "text" : texts, }
pass

from datasets import Dataset
dataset = Dataset.from_pandas(formatted_df, split = 'train')
dataset = dataset.map(formatting_prompts_func, batched = True,)
print(dataset)

from trl import SFTTrainer
from transformers import TrainingArguments
from unsloth import is_bfloat16_supported

trainer = SFTTrainer(
    model = model,
    tokenizer = tokenizer,
    train_dataset = dataset,
    dataset_text_field = "text",
    max_seq_length = max_seq_length,
    dataset_num_proc = 3,
    packing = False, # Can make training 5x faster for short sequences.
    args = TrainingArguments(
        per_device_train_batch_size = 2,
        gradient_accumulation_steps = 4,
        warmup_steps = 5,
        max_steps = 60,
        learning_rate = 2e-4,
        fp16 = not is_bfloat16_supported(),
        bf16 = is_bfloat16_supported(),
        logging_steps = 1,
        optim = "adamw_8bit",
        weight_decay = 0.01,
        lr_scheduler_type = "linear",
        seed = 3407,
        output_dir = "outputs",
    ),
)

#@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.")

trainer_stats = trainer.train()

#@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} %.")

"""<a name="Inference"></a>
### Inference
Let's run the model! You can change the instruction and input - leave the output blank!

You can also use a `TextStreamer` for continuous inference - so you can see the generation token by token, instead of waiting the whole time!
"""

from transformers import TextStreamer
FastLanguageModel.for_inference(model) # Enable native 2x faster inference

import json
import re

def extract_json(text):
    # Regular expression pattern to match JSON objects
    json_pattern = re.compile(r'\{.*?\}', re.DOTALL)
    potential_jsons = json_pattern.findall(text)
    extracted_jsons = []
    for potential_json in potential_jsons:
        try:
            extracted_jsons.append(json.loads(potential_json))
        except json.JSONDecodeError:
            continue
    return extracted_jsons[0:1]

# alpaca_prompt = You MUST copy from above!
inputs = tokenizer(
[
    alpaca_prompt.format(
        "Find the derivative of f(x) = 3x^2 + 4cos(x) - 1 for a maximum of 2 marks.'. Referring to the mark-scheme, award the appropriate amount of marks to the student: 'Correctly apply differentiation rules.", # instruction
        "6x^2 - 4sin(x)", # input
        "", # output - leave this blank for generation!
    )
], return_tensors = "pt").to("cuda")

outputs = model.generate(**inputs, max_new_tokens = 64, use_cache = True)
tokenizer.batch_decode(outputs)