Upload 4 files

App.py

@@ -0,0 +1,105 @@
+
+import streamlit as st
+import pandas as pd
+from datetime import datetime, timedelta
+from unsloth import FastLanguageModel
+import torch
+
+# Cargar el modelo y el tokenizador
+model_path = "/home/roser97/MarketAI/lora_model"
+model, tokenizer = FastLanguageModel.from_pretrained(
+    model_name=model_path,
+    max_seq_length=800,  # Ajusta según tus necesidades
+    load_in_4bit=True,
+)
+
+# Configurar el modelo para inferencia
+FastLanguageModel.for_inference(model)
+
+def generate_marketing_content(instruction, input_context):
+    inputs = tokenizer(
+        [f"### Instruction:\n{instruction}\n### Input:\n{input_context}\n### Response:"],
+        return_tensors="pt"
+    ).to("cuda" if torch.cuda.is_available() else "cpu")
+
+    output = model.generate(**inputs, max_new_tokens=128)
+    return tokenizer.decode(output[0], skip_special_tokens=True)
+
+def main():
+    st.set_page_config(page_title="Compass AI", layout="wide")
+    st.title("Compass AI")
+
+    # Sidebar for navigation
+    page = st.sidebar.selectbox("Choose a page", ["Home", "Campaign Creation", "Strategy", "Scheduling", "Analytics"])
+
+    if page == "Home":
+        show_home()
+    elif page == "Campaign Creation":
+        show_campaign_creation()
+    elif page == "Strategy":
+        show_strategy()
+    elif page == "Scheduling":
+        show_scheduling()
+    elif page == "Analytics":
+        show_analytics()
+
+def show_home():
+    st.header("Welcome to AI Marketing Campaign Agent")
+    st.write("This tool helps you create, manage, and analyze your marketing campaigns using AI.")
+    st.write("Use the sidebar to navigate through different features.")
+
+def show_campaign_creation():
+    st.header("Campaign Creation")
+    
+    # Brand Questionnaire
+    st.subheader("Brand Questionnaire")
+    brand_name = st.text_input("Brand Name")
+    industry = st.selectbox("Industry", ["Technology", "Fashion", "Food & Beverage", "Other"])
+    target_audience = st.text_area("Describe your target audience")
+    campaign_objective = st.selectbox("Campaign Objective", ["Brand Awareness", "Lead Generation", "Sales", "Other"])
+
+    # Content Generation
+    st.subheader("Content Generation")
+    content_type = st.selectbox("Content Type", ["Social Media Post", "Ad Copy", "Email"])
+    content_prompt = st.text_area("Describe the content you want to generate")
+
+    if st.button("Generate Content"):
+        with st.spinner("Generating content..."):
+            generated_content = generate_marketing_content(content_prompt, f"{brand_name}, {industry}, {target_audience}, {campaign_objective}")
+        st.text_area("Generated Content", generated_content, height=200)
+
+def show_strategy():
+    st.header("Marketing Strategy")
+    
+    start_date = st.date_input("Campaign Start Date")
+    duration = st.number_input("Campaign Duration (days)", min_value=1, value=30)
+    
+    if st.button("Generate Strategy"):
+        with st.spinner("Generating strategy..."):
+            strategy = generate_marketing_content("Generate a marketing strategy", f"Start Date: {start_date}, Duration: {duration} days")
+        
+        st.subheader("Generated Marketing Strategy")
+        st.text(strategy)
+        
+        if st.button("Generate PDF Proposal"):
+            st.write("PDF generation functionality to be implemented.")
+
+def show_scheduling():
+    st.header("Content Scheduling")
+    
+    platforms = st.multiselect("Select Platforms", ["Facebook", "Instagram", "Twitter"])
+    post_content = st.text_area("Post Content")
+    post_date = st.date_input("Post Date")
+    post_time = st.time_input("Post Time")
+    
+    if st.button("Schedule Post"):
+        scheduled_datetime = datetime.combine(post_date, post_time)
+        for platform in platforms:
+            st.success(f"Post scheduled for {platform} at {scheduled_datetime}")
+
+def show_analytics():
+    st.header("Campaign Analytics")
+    st.write("This feature is under development. It will show campaign performance metrics and insights.")
+    
+if __name__ == "__main__":
+    main()

Train.py

@@ -0,0 +1,219 @@
+# -*- coding: utf-8 -*-
+"""train.py.ipynb
+
+Automatically generated by Colab.
+
+Original file is located at
+    https://colab.research.google.com/drive/1KS0vOkR4vSJYwCiFMW4o9c8RK1h5hpCt
+"""
+
+!pip install "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
+!pip install --no-deps "xformers<0.0.27" "trl<0.9.0" peft accelerate bitsandbytes  ratelimit
+!pip install torch transformers datasets rich
+
+from unsloth import FastLanguageModel
+import torch
+
+# Configuraciones del modelo
+max_seq_length = 800  # Ajusta según tus necesidades
+dtype = None  # None para detección automática. Usa Float16 o Bfloat16 si sabes cuál usar.
+load_in_4bit = True  # Habilita la cuantificación de 4 bits para reducir el uso de memoria
+
+# Carga el modelo preentrenado con las optimizaciones
+model, tokenizer = FastLanguageModel.from_pretrained(
+    model_name="unsloth/Meta-Llama-3.1-8B",  # Nombre del modelo preentrenado
+    max_seq_length=max_seq_length,  # Longitud máxima de secuencia
+    dtype=dtype,  # Tipo de dato de la GPU (detección automática si es None)
+    load_in_4bit=load_in_4bit,  # Cuantificación de 4 bits para optimizar memoria
+)
+
+print("Modelo cargado exitosamente.")
+
+# Aplicación de adaptadores LoRA al modelo
+model = FastLanguageModel.get_peft_model(
+    model,
+    r=16,
+    target_modules=["q_proj", "k_proj", "v_proj", "o_proj", "gate_proj", "up_proj", "down_proj"],
+    lora_alpha=16,
+    lora_dropout=0,
+    bias="none",
+    use_gradient_checkpointing="unsloth",
+    random_state=3407,
+    use_rslora=False,
+    loftq_config=None,
+)
+
+print("Adaptadores LoRA aplicados correctamente.")
+
+from datasets import load_dataset
+
+# Paso 8: Dar formato al conjunto de datos para el entrenamiento
+# Definir la plantilla de formato
+alpaca_prompt = """Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.
+
+### Instruction:
+{}
+
+### Input:
+{}
+
+### Response:
+{}"""
+EOS_TOKEN = tokenizer.eos_token
+
+# Función para formatear el conjunto de datos
+def formatting_prompts_func(examples):
+    instructions = examples["instruction"]
+    inputs = examples["input"]
+    outputs = examples["response"]
+    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}
+
+# Cargar y preparar el dataset para el ajuste fino
+dataset = load_dataset('json', data_files='/content/marketing_social_media_dataset_v1.json', split='train')
+
+# Aplicar el formato a todo el dataset usando la función map
+dataset = dataset.map(formatting_prompts_func, batched=True)
+
+print("Dataset cargado y formateado correctamente.")
+
+from trl import SFTTrainer
+from transformers import TrainingArguments
+
+# Configuración del entrenamiento utilizando SFTTrainer
+trainer = SFTTrainer(
+    model=model,
+    tokenizer=tokenizer,
+    train_dataset=dataset,
+    dataset_text_field="text",
+    max_seq_length=max_seq_length,
+    dataset_num_proc=2,
+    packing=False,
+    args=TrainingArguments(
+        per_device_train_batch_size=2,
+        gradient_accumulation_steps=4,
+        warmup_steps=5,
+        max_steps=60,
+        learning_rate=2e-4,
+        fp16=True,  # Deshabilitar fp16
+        bf16=False,   # Habilitar bf16, recomendado para GPUs Ampere
+        logging_steps=1,
+        optim="adamw_8bit",
+        weight_decay=0.01,
+        lr_scheduler_type="linear",
+        seed=3407,
+        output_dir="outputs",
+    ),
+)
+
+# Iniciar el proceso de entrenamiento
+trainer_stats = trainer.train()
+
+print("Entrenamiento completado con éxito.")
+
+from rich.console import Console
+from rich.panel import Panel
+from rich.text import Text
+from rich.markdown import Markdown as RichMarkdown
+from IPython.display import display, Markdown
+import json
+
+# Configurar el modelo para la inferencia
+FastLanguageModel.for_inference(model)
+
+# Generar texto basado en una instrucción dada
+inputs = tokenizer(
+    [
+        alpaca_prompt.format(
+            "Best marketing post for sneaker company",  # Instrucción para el modelo
+            "",  # Entrada adicional (en este caso, ninguna)
+            "",  # Respuesta esperada (en este caso, ninguna)
+        )
+    ], return_tensors="pt").to("cuda")
+
+# Generar la salida sin el uso de TextStreamer
+output = model.generate(**inputs)
+
+# Decodificar la salida
+output_text = tokenizer.decode(output[0], skip_special_tokens=True)
+
+# Función para analizar la salida y convertirla en un diccionario
+def parse_output_to_dict(output_text):
+    result = {}
+    current_section = None
+    lines = output_text.split('\n')
+    for line in lines:
+        line = line.strip()
+        if line.startswith('###'):
+            current_section = line.strip('# ').lower().replace(' ', '_')
+            result[current_section] = {}
+        elif ':' in line:
+            key, value = line.split(':', 1)
+            key = key.lower().replace(' ', '_').strip()
+            result[current_section][key] = value.strip()
+        elif line and current_section:
+            if 'content' not in result[current_section]:
+                result[current_section]['content'] = []
+            result[current_section]['content'].append(line)
+
+    return result
+
+# Analizar la salida generada en un diccionario
+parsed_output = parse_output_to_dict(output_text)
+
+# Mostrar la salida analizada como un JSON formateado
+display(Markdown("## Parsed JSON Output\n\n```json\n" + json.dumps(parsed_output, indent=2) + "\n```"))
+
+# Guardar el modelo ajustado y el tokenizador en un directorio
+model.save_pretrained("lora_model")
+tokenizer.save_pretrained("lora_model")
+
+print("Modelo y tokenizador guardados correctamente en 'lora_model'.")
+
+from unsloth import FastLanguageModel
+
+# Recargar el modelo y el tokenizador desde el directorio guardado
+model, tokenizer = FastLanguageModel.from_pretrained(
+    model_name="lora_model",
+    max_seq_length=max_seq_length,
+    dtype=dtype,
+    load_in_4bit=load_in_4bit,
+)
+
+# Configurar el modelo para inferencia
+FastLanguageModel.for_inference(model)
+
+print("Modelo y tokenizador recargados correctamente desde 'lora_model'.")
+
+# Generar texto basado en una nueva indicación
+inputs = tokenizer(
+    [
+        alpaca_prompt.format(
+            "Create a marketing campaign to promote the chocolate bar",  # Instrucción
+            "Company: Cadbury, target audience: adults/boomers",  # Información de entrada adicional
+            "",  # Respuesta esperada (en este caso, ninguna)
+        )
+    ], return_tensors="pt").to("cuda")
+
+# Generar salida (si no usas TextStreamer, puedes quitar la línea correspondiente)
+output = model.generate(**inputs, max_new_tokens=128)
+
+# Decodificar la salida
+output_text = tokenizer.decode(output[0], skip_special_tokens=True)
+
+# Mostrar la salida generada
+print("Salida Generada por el Modelo:")
+print(output_text)
+
+from google.colab import files
+import shutil
+
+# Crear un archivo zip del modelo
+shutil.make_archive("lora_model", 'zip', "lora_model")
+
+# Descargar el archivo zip
+files.download("lora_model.zip")
+

requirements (1).txt

@@ -0,0 +1,5 @@
+pip install "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
+pip install --no-deps "xformers<0.0.27" "trl<0.9.0" peft accelerate bitsandbytes  ratelimit
+pip install torch transformers datasets rich
+ipython
+openai

requirements.txt

@@ -0,0 +1,117 @@
+accelerate==0.33.0
+aiohappyeyeballs==2.4.0
+aiohttp==3.10.4
+aiosignal==1.3.1
+altair==5.4.0
+annotated-types==0.7.0
+anyio==4.4.0
+asttokens==2.4.1
+async-timeout==4.0.3
+attrs==24.2.0
+bitsandbytes==0.43.3
+blinker==1.8.2
+cachetools==5.5.0
+certifi==2024.7.4
+charset-normalizer==3.3.2
+click==8.1.7
+datasets==2.21.0
+decorator==5.1.1
+dill==0.3.8
+distro==1.9.0
+docstring_parser==0.16
+exceptiongroup==1.2.2
+executing==2.0.1
+filelock==3.15.4
+frozenlist==1.4.1
+fsspec==2024.6.1
+gitdb==4.0.11
+GitPython==3.1.43
+h11==0.14.0
+hf_transfer==0.1.8
+httpcore==1.0.5
+httpx==0.27.0
+huggingface-hub==0.24.5
+idna==3.7
+ipython==8.26.0
+jedi==0.19.1
+Jinja2==3.1.4
+jiter==0.5.0
+jsonschema==4.23.0
+jsonschema-specifications==2023.12.1
+markdown-it-py==3.0.0
+MarkupSafe==2.1.5
+matplotlib-inline==0.1.7
+mdurl==0.1.2
+mpmath==1.3.0
+multidict==6.0.5
+multiprocess==0.70.16
+narwhals==1.4.2
+networkx==3.3
+numpy==2.0.1
+nvidia-cublas-cu12==12.1.3.1
+nvidia-cuda-cupti-cu12==12.1.105
+nvidia-cuda-nvrtc-cu12==12.1.105
+nvidia-cuda-runtime-cu12==12.1.105
+nvidia-cudnn-cu12==9.1.0.70
+nvidia-cufft-cu12==11.0.2.54
+nvidia-curand-cu12==10.3.2.106
+nvidia-cusolver-cu12==11.4.5.107
+nvidia-cusparse-cu12==12.1.0.106
+nvidia-nccl-cu12==2.20.5
+nvidia-nvjitlink-cu12==12.6.20
+nvidia-nvtx-cu12==12.1.105
+openai==1.41.0
+packaging==24.1
+pandas==2.2.2
+parso==0.8.4
+peft==0.12.0
+pexpect==4.9.0
+pillow==10.4.0
+prompt_toolkit==3.0.47
+protobuf==3.20.3
+psutil==6.0.0
+ptyprocess==0.7.0
+pure_eval==0.2.3
+pyarrow==17.0.0
+pydantic==2.8.2
+pydantic_core==2.20.1
+pydeck==0.9.1
+Pygments==2.18.0
+python-dateutil==2.9.0.post0
+pytz==2024.1
+PyYAML==6.0.2
+ratelimit==2.2.1
+referencing==0.35.1
+regex==2024.7.24
+requests==2.32.3
+rich==13.7.1
+rpds-py==0.20.0
+safetensors==0.4.4
+sentencepiece==0.2.0
+shtab==1.7.1
+six==1.16.0
+smmap==5.0.1
+sniffio==1.3.1
+stack-data==0.6.3
+streamlit==1.37.1
+sympy==1.13.2
+tenacity==8.5.0
+tokenizers==0.19.1
+toml==0.10.2
+torch==2.4.0
+tornado==6.4.1
+tqdm==4.66.5
+traitlets==5.14.3
+transformers==4.44.0
+triton==3.0.0
+trl==0.9
+typing_extensions==4.12.2
+tyro==0.8.8
+tzdata==2024.1
+unsloth @ git+https://github.com/unslothai/unsloth.git@52bc19d1fa4cd3557b785127fd68b5f4d1c34347
+urllib3==2.2.2
+watchdog==4.0.2
+wcwidth==0.2.13
+xformers==0.0.27
+xxhash==3.5.0
+yarl==1.9.4