ضبط درجة الحرارة (Temperature)

import csv
import string
import random
import math
from collections import defaultdict

with open("slm-corpus.csv", newline="") as f:
    reader = csv.DictReader(f)
    texts = [row["text"] for row in reader]

def load_corpus(path):
    with open(path, newline="") as f:
        reader = csv.DictReader(f)
        return [row["text"] for row in reader]

def tokenize(text):
    text = text.lower()
    for char in string.punctuation:
        text = text.replace(char, " ")
    return text.split()

def build_bigrams(tokens):
    bigrams = defaultdict(lambda: defaultdict(int))
    for i in range(len(tokens) - 1):
        bigrams[tokens[i]][tokens[i + 1]] += 1
    return dict(bigrams)

def normalize_bigrams(bigrams):
    normalized = {}
    for word, followers in bigrams.items():
        if not followers:
            continue
        total = sum(followers.values())
        normalized[word] = {w: c / total for w, c in followers.items()}
    return normalized

def apply_temperature(probs, temperature):
    log_probs = [math.log(p + 1e-10) for p in probs]
    scaled = [lp / temperature for lp in log_probs]
    max_s = max(scaled)
    exp_s = [math.exp(s - max_s) for s in scaled]
    total = sum(exp_s)
    return [e / total for e in exp_s]

def sample_next(model, current_word, temperature=1.0):
    if current_word not in model:
        return None
    followers = model[current_word]
    words = list(followers.keys())
    probs = list(followers.values())
    if temperature != 1.0:
        probs = apply_temperature(probs, temperature)
    return random.choices(words, weights=probs, k=1)[0]

def generate_text(model, start_word, length=20, temperature=1.0):
    word = start_word
    result = [word]
    for _ in range(length - 1):
        next_word = sample_next(model, word, temperature)
        if next_word is None:
            next_word = random.choice(["the", "and", "to", "of", "a"])
        result.append(next_word)
        word = next_word
    return " ".join(result)

model = normalize_bigrams(build_bigrams(tokenize(" ".join(texts))))

import math

def apply_temperature(probabilities, temperature):
    """Apply temperature scaling to a probability distribution."""
    # Convert to log-probabilities
    log_probs = [math.log(p + 1e-10) for p in probabilities]  # add small epsilon to avoid log(0)

    # Scale by temperature
    scaled = [lp / temperature for lp in log_probs]

    # Convert back to probabilities (softmax-like)
    max_scaled = max(scaled)
    exp_scaled = [math.exp(s - max_scaled) for s in scaled]  # subtract max for numerical stability
    total = sum(exp_scaled)

    return [e / total for e in exp_scaled]

words = ["cat", "dog", "bird"]
probs = [0.7, 0.2, 0.1]

# Low temperature: cat becomes even more dominant
cold = apply_temperature(probs, temperature=0.5)
print("Cold (0.5):", dict(zip(words, [f"{p:.3f}" for p in cold])))
# cat ≈ 0.876, dog ≈ 0.088, bird ≈ 0.036

# High temperature: more uniform distribution
hot = apply_temperature(probs, temperature=2.0)
print("Hot (2.0):", dict(zip(words, [f"{p:.3f}" for p in hot])))
# cat ≈ 0.524, dog ≈ 0.281, bird ≈ 0.195

import random

def sample_next(model, current_word, temperature=1.0):
    if current_word not in model:
        return None

    followers = model[current_word]
    words = list(followers.keys())
    probs = list(followers.values())

    if temperature != 1.0:
        probs = apply_temperature(probs, temperature)

    return random.choices(words, weights=probs, k=1)[0]

# Cold: repetitive, predictable
random.seed(42)
for _ in range(3):
    print(generate_text(model, "the", length=10, temperature=0.3))

# Hot: creative, surprising
random.seed(42)
for _ in range(3):
    print(generate_text(model, "the", length=10, temperature=1.5))

random.seed(42)
for temp in [0.3, 1.0, 1.5]:
    print(f"\n[temperature={temp}]")
    for _ in range(3):
        print(f"  {generate_text(model, 'the', length=12, temperature=temp)}")

def compare_temperatures(model, word, temps=[0.3, 0.7, 1.0, 1.5], length=15):
    for temp in temps:
        random.seed(42)
        text = generate_text(model, word, length=length, temperature=temp)
        print(f"  T={temp:.1f}: {text}")