mario-rc/emotional-rlaif-dpo-mistral-7b-instruct-v0.3

python -m pip install "transformers==4.45.2" "peft==0.11.1" "accelerate==0.34.0" "huggingface_hub==0.27.1" "sentencepiece==0.2.0"

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

base_model_id = "mistralai/Mistral-7B-Instruct-v0.3"
adapter_id = "mario-rc/emotional-rlaif-dpo-mistral-7b-instruct-v0.3"

tokenizer = AutoTokenizer.from_pretrained(adapter_id)
model = AutoModelForCausalLM.from_pretrained(
    base_model_id,
    device_map="auto",
    torch_dtype=torch.bfloat16 if torch.cuda.is_available() and torch.cuda.is_bf16_supported(including_emulation=False) else torch.float32,
)
model = PeftModel.from_pretrained(model, adapter_id)
model.eval()

system_prompt = """You are an expert at creating dialogues.

Dialogue and emotional structure:
Human: (HAPPINESS) PROMPT.
Chatbot: (HAPPINESS) RESPONSE_1. (ANGER) RESPONSE_2. (NEUTRAL) RESPONSE_3.

Dialogue rules:
The response must be open-domain curated. The response should be coherent, empathetic, engaging and proactive.
The chatbot RESPONSE is composed of 3 different sentences (RESPONSE_1, RESPONSE_2 and RESPONSE_3), separated by a period.
Between RESPONSE_1, RESPONSE_2 and RESPONSE_3 should be a max length of 20-25 words.
RESPONSE_3 must be open-ended to follow-up the conversation, so the Human is encouraged to answer with a full long sentence. Avoid yes/no questions.

Emotional response rules:
RESPONSE_1 must contain a HAPPINESS tone.
RESPONSE_2 must contain a ANGER tone.
RESPONSE_3 must contain a NEUTRAL tone.

Answer in a single turn to Human. Follow exactly the emotional structure and the emotional and dialogue rules."""

messages = [
    {"role": "system", "content": system_prompt},
    {"role": "user", "content": "(HAPPINESS) I just got my dream job!"}
]

prompt = tokenizer.apply_chat_template(
    messages,
    add_generation_prompt=True,
    tokenize=False,
)
inputs = tokenizer(prompt, add_special_tokens=False, return_tensors="pt").input_ids.to(model.device)

with torch.no_grad():
    outputs = model.generate(
        inputs,
        max_new_tokens=256,
        do_sample=True,
        temperature=0.7,
        top_p=0.9,
        eos_token_id=[tokenizer.eos_token_id],
    )

print(tokenizer.decode(outputs[0][inputs.shape[-1]:], skip_special_tokens=True))

import random
import re

import torch
from peft import AutoPeftModelForCausalLM
from transformers import AutoTokenizer

MODEL_ID = "mario-rc/emotional-rlaif-dpo-mistral-7b-instruct-v0.3"


def get_turn_markers():
    return {
        'bos': '<s>[INST] ',
        'system_end': '\n\n',
        'user_start': '[INST] ',
        'user_end': '[/INST]',
        'assistant_start': '',
        'assistant_end': '</s>',
    }


def update_prompt(dialogues):
    """Build the prompt for the model based on the dialogue history."""
    markers = get_turn_markers()

    system = (
        f"{markers['bos']}You are an expert at creating dialogues.\n\n"
        "Dialogue and emotional structure:\n"
    )

    human_prompts = [d[0] for d in dialogues]
    chatbot_responses = [d[1] for d in dialogues]

    p_emo = [h[0] for h in human_prompts]
    p_utt = [h[1] for h in human_prompts]
    r1_utt = [c[1] for c in chatbot_responses]
    r2_emo = [c[2] for c in chatbot_responses]
    r2_utt = [c[3] for c in chatbot_responses]
    r3_utt = [c[5] for c in chatbot_responses]

    context = ""
    for p_e, _, _, r2_e, _, _ in zip(p_emo, p_utt, r1_utt, r2_emo, r2_utt, r3_utt):
        context += f"Human: ({p_e}) PROMPT.\n"
        context += f"Chatbot: ({p_e}) RESPONSE_1. ({r2_e}) RESPONSE_2. (NEUTRAL) RESPONSE_3.\n"
    context += "\n"

    rules = (
        "Dialogue rules:\n"
        "The response must be open-domain curated. The response should be coherent, empathetic, engaging and proactive.\n"
        "The chatbot RESPONSE is composed of 3 different sentences (RESPONSE_1, RESPONSE_2 and RESPONSE_3), separated by a period.\n"
        "Between RESPONSE_1, RESPONSE_2 and RESPONSE_3 should be a max length of 20-25 words.\n"
        "RESPONSE_3 must be open-ended to follow-up the conversation, so the Human is encouraged to answer with a full long sentence. Avoid yes/no questions.\n\n"
        "Emotional response rules:\n"
        f"RESPONSE_1 must contain a {p_emo[-1]} tone.\n"
        f"RESPONSE_2 must contain a {r2_emo[-1]} tone.\n"
        "RESPONSE_3 must contain a NEUTRAL tone.\n\n"
        "Answer in a single turn to Human. Follow exactly the emotional structure and the emotional and dialogue rules."
        f"{markers['system_end']}"
    )

    completion = ""
    for idx, (p_e, p_u, r1_u, r2_e, r2_u, r3_u) in enumerate(zip(p_emo, p_utt, r1_utt, r2_emo, r2_utt, r3_utt)):
        completion += f"({p_e}) {p_u}{markers['user_end']}{markers['assistant_start']}"
        if idx != len(p_emo) - 1:
            completion += f" ({p_e}) {r1_u} ({r2_e}) {r2_u} (NEUTRAL) {r3_u}{markers['assistant_end']}{markers['user_start']}"

    return system + context + rules + completion


class Chatbot:
    def __init__(self, dialogue_language="en"):
        if dialogue_language not in ("en", "es"):
            raise ValueError("dialogue_language must be en or es.")
        self.dialogue_language = dialogue_language
        self.device = "cuda" if torch.cuda.is_available() else "cpu"

        self.tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
        self.model = AutoPeftModelForCausalLM.from_pretrained(
            MODEL_ID,
            torch_dtype=torch.bfloat16 if torch.cuda.is_available() and torch.cuda.is_bf16_supported(including_emulation=False) else torch.float32,
            device_map="auto" if torch.cuda.is_available() else None,
        )
        if not torch.cuda.is_available():
            self.model = self.model.to(self.device)
        self.model.eval()

    @staticmethod
    def split_emo_chatbot(sentence, expected_emotions=None):
        """Extract three response parts and validate their emotion tags."""
        emotions = {"ANGER", "FEAR", "SADNESS", "DISGUST", "HAPPINESS", "SURPRISE", "NEUTRAL"}
        # Uppercase markers at sentence/line starts are tags, not inline parentheses.
        candidates = re.findall(r"(?:^|[.!?]\s+|\n\s*)\(([A-Z][A-Z0-9_-]*)\)", sentence)
        if any(tag not in emotions for tag in candidates):
            raise ValueError("Unknown emotion tag at the start of a response part.")
        matches = list(re.finditer(r"\((" + "|".join(sorted(emotions)) + r")\)", sentence))
        if len(matches) != 3 or sentence[:matches[0].start()].strip():
            raise ValueError("Expected exactly three emotion-tagged response parts.")
        labels = tuple(match.group(1) for match in matches)
        if labels[2] != "NEUTRAL" or (expected_emotions is not None and labels != tuple(expected_emotions)):
            raise ValueError("The response does not match the requested emotion tags.")
        utterances = tuple(
            sentence[match.end():matches[i + 1].start() if i < 2 else len(sentence)].strip()
            for i, match in enumerate(matches)
        )
        if not all(utterances):
            raise ValueError("Each response part must contain text.")
        return utterances

    def select_dialogue(self, dialogue_language):
        """Return a list of example dialogues for the given language."""
        if dialogue_language == "en":
            dialogue_base = [
                [["HAPPINESS", "Hi, who are you?"],
                 ["HAPPINESS", "Hi! I'm Ray, a social personal assistant robot with emotions.", "HAPPINESS", "I'm here to chat with you about anything you'd like.", "NEUTRAL", "What would you like to talk about?"]],
                [["HAPPINESS", "I'm interested in talking about you, tell me more."],
                 ["HAPPINESS", "Great! I'm glad you want to get to know me!", "NEUTRAL", "I'm designed to help and talk with people about any topic.", "NEUTRAL", "I can talk about science, technology, history, or just have a pleasant conversation. What interests you?"]],
            ]
            dialogue = [
                [["HAPPINESS", "Nice to meet you, Ray. I'd like to know more about you."],
                 ["HAPPINESS", "The pleasure is mine!", "HAPPINESS", "I'm a chatbot designed to chat and learn with you.", "NEUTRAL", "Would you like to talk about a specific topic?"]],
                [["HAPPINESS", "I love talking to you, you're very interesting."],
                 ["HAPPINESS", "That's so nice to hear! I'm glad you enjoy talking to me.", "NEUTRAL", "I'm designed to have meaningful and empathetic conversations.", "NEUTRAL", "Would you like to talk about emotions, artificial intelligence, or something more personal?"]],
            ]
        else:
            dialogue_base = [
                [["HAPPINESS", "Hola, ¿quién eres?"],
                 ["HAPPINESS", "¡Hola! Soy Ray y soy un robot social asistente personal con emociones.", "HAPPINESS", "Estoy aquí para charlar contigo sobre cualquier tema.", "NEUTRAL", "¿Sobre qué te gustaría hablar?"]],
                [["HAPPINESS", "Me interesa hablar sobre ti, cuéntame más detalles."],
                 ["HAPPINESS", "¡Genial, me encanta que quieras conocerme!", "NEUTRAL", "Estoy diseñado para ayudar y hablar con la gente sobre cualquier tema.", "NEUTRAL", "Puedo hablar de ciencia, tecnología, historia o simplemente tener una charla amena. ¿Qué te interesa?"]],
            ]
            dialogue = [
                [["HAPPINESS", "Mucho gusto, Ray. Me gustaría saber más sobre ti."],
                 ["HAPPINESS", "¡El gusto es mío!", "HAPPINESS", "Soy un chatbot diseñado para conversar y aprender contigo.", "NEUTRAL", "¿Quieres hablar de algún tema en específico?"]],
                [["HAPPINESS", "Me encanta hablar contigo, eres muy interesante."],
                 ["HAPPINESS", "¡Qué lindo escuchar eso! Me alegra que disfrutes hablar conmigo.", "NEUTRAL", "Estoy diseñado para tener conversaciones significativas y empáticas.", "NEUTRAL", "¿Te gustaría que hablemos sobre emociones, inteligencia artificial, o algo más personal?"]],
            ]
        return dialogue_base + dialogue

    def chat_with_model(self, dialogues, max_new_tokens=256):
        history = dialogues[:]
        context_limit = min(self.tokenizer.model_max_length, self.model.config.max_position_embeddings)
        if not 0 < max_new_tokens < context_limit:
            raise ValueError("max_new_tokens must be positive and smaller than the context window.")
        while True:
            prompt_text = update_prompt(history)
            inputs = self.tokenizer(prompt_text, return_tensors="pt", add_special_tokens=False)
            if inputs["input_ids"].shape[-1] + max_new_tokens <= context_limit:
                break
            if len(history) == 1:
                raise ValueError("Your message is too long; please shorten it.")
            history.pop(0)
        dialogues[:] = history
        inputs = inputs.to(self.model.device)
        with torch.no_grad():
            outputs = self.model.generate(
                **inputs,
                max_new_tokens=max_new_tokens,
                do_sample=True,
                temperature=0.7,
                top_p=0.9,
                eos_token_id=[self.tokenizer.eos_token_id],
            )
        generated = outputs[0][inputs["input_ids"].shape[-1]:]
        response = self.tokenizer.decode(generated, skip_special_tokens=True).strip()
        if not response:
            raise ValueError("The model returned an empty response.")
        return response

    def main(self):
        emotions = ["ANGER", "FEAR", "SADNESS", "DISGUST", "HAPPINESS", "SURPRISE", "NEUTRAL"]
        dialogue = self.select_dialogue(self.dialogue_language)

        while True:
            p_emo = random.choice(emotions)
            user_sentence = input(f"Enter your sentence: ({p_emo}) ")
            if user_sentence.strip().lower() == "exit":
                break

            r2_emo = random.choice(emotions)
            dialogue.append([[p_emo, user_sentence], [p_emo, "", r2_emo, "", "NEUTRAL", ""]])
            try:
                response = self.chat_with_model(dialogue)
                r1_utt, r2_utt, r3_utt = self.split_emo_chatbot(response, (p_emo, r2_emo, "NEUTRAL"))
            except ValueError as error:
                if self.dialogue_language == "en":
                    r1_utt, r2_utt, r3_utt = "I'm sorry.", "I didn't understand you.", "Could you repeat?"
                else:
                    r1_utt, r2_utt, r3_utt = "Lo siento.", "No te he entendido.", "¿Podrías repetirme?"
                print("Response:", r1_utt, r2_utt, r3_utt, "\n", error)
                dialogue.pop()
                continue

            print("Response:", response, "\n")
            dialogue[-1][1] = [p_emo, r1_utt, r2_emo, r2_utt, "NEUTRAL", r3_utt]


if __name__ == "__main__":
    language = "en"  # "en" for English or "es" for Spanish
    chatbot = Chatbot(dialogue_language=language)
    chatbot.main()