eval_BLEU.py

import os
import pickle
from contextlib import nullcontext
import torch
import tiktoken
from model import GPTConfig, GPT
import sacrebleu
import pandas as pd

# -----------------------------------------------------------------------------
init_from = 'gpt2'
#init_from = 'out-arc' # either 'resume' (from an out_dir) or a gpt2 variant (e.g. 'gpt2-xl')
out_dir = 'out' # ignored if init_from is not 'resume'
start = "\n" # or "<|endoftext|>" or etc. Can also specify a file, use as: "FILE:prompt.txt"


temperature = 0.8 # 1.0 = no change, < 1.0 = less random, > 1.0 = more random, in predictions
top_k = 200 # retain only the top_k most likely tokens, clamp others to have 0 probability
seed = 1337
device = 'cuda' # examples: 'cpu', 'cuda', 'cuda:0', 'cuda:1', etc.
dtype = 'bfloat16' if torch.cuda.is_available() and torch.cuda.is_bf16_supported() else 'float16' # 'float32' or 'bfloat16' or 'float16'
compile = False # use PyTorch 2.0 to compile the model to be faster

sparse = 0.


exec(open('configurator.py').read()) # overrides from command line or config file


# -----------------------------------------------------------------------------

torch.manual_seed(seed)
torch.cuda.manual_seed(seed)
torch.backends.cuda.matmul.allow_tf32 = True # allow tf32 on matmul
torch.backends.cudnn.allow_tf32 = True # allow tf32 on cudnn
device_type = 'cuda' if 'cuda' in device else 'cpu' # for later use in torch.autocast
ptdtype = {'float32': torch.float32, 'bfloat16': torch.bfloat16, 'float16': torch.float16}[dtype]
ctx = nullcontext() if device_type == 'cpu' else torch.amp.autocast(device_type=device_type, dtype=ptdtype)

# model
if init_from == 'resume':
    # init from a model saved in a specific directory
    ckpt_path = os.path.join(out_dir, 'ckpt.pt')
    checkpoint = torch.load(ckpt_path, map_location=device)
    gptconf = GPTConfig(**checkpoint['model_args'])
    model = GPT(gptconf)
    state_dict = checkpoint['model']
    unwanted_prefix = '_orig_mod.'
    for k,v in list(state_dict.items()):
        if k.startswith(unwanted_prefix):
            state_dict[k[len(unwanted_prefix):]] = state_dict.pop(k)
    model.load_state_dict(state_dict)
elif init_from.startswith('gpt2'):
    # init from a given GPT-2 model
    model = GPT.from_pretrained(init_from, dict(dropout=0.0))



model.eval()
model.to(device)
if compile:
    model = torch.compile(model) # requires PyTorch 2.0 (optional)

# look for the meta pickle in case it is available in the dataset folder
load_meta = False
if init_from == 'resume' and 'config' in checkpoint and 'dataset' in checkpoint['config']: # older checkpoints might not have these...
    meta_path = os.path.join('data', checkpoint['config']['dataset'], 'meta.pkl')
    load_meta = os.path.exists(meta_path)
if load_meta:
    print(f"Loading meta from {meta_path}...")
    with open(meta_path, 'rb') as f:
        meta = pickle.load(f)
    # TODO want to make this more general to arbitrary encoder/decoder schemes
    stoi, itos = meta['stoi'], meta['itos']
    encode = lambda s: [stoi[c] for c in s]
    decode = lambda l: ''.join([itos[i] for i in l])
else:
    # ok let's assume gpt-2 encodings by default
    print("No meta.pkl found, assuming GPT-2 encodings...")
    enc = tiktoken.get_encoding("gpt2")
    encode = lambda s: enc.encode(s, allowed_special={"<|endoftext|>"})
    decode = lambda l: enc.decode(l)






def ans(question):
    question = "Translate this English sentence to French and only include the French sentence and no English words:" +question
    start_id = encode(question)
    x = (torch.tensor(start_id, dtype=torch.long, device=device)[None, ...])
    with torch.no_grad():
        with ctx:
            y = model.generate(x, 1, temperature=temperature, top_k=top_k)
    answer = decode(y[0].tolist())
    return answer.upper()


def getscoreBLEU(question, answer):
    ref = [[answer]]
    sys = [ans(question)]

    bleu = sacrebleu.corpus_bleu(sys, ref)
    return bleu.score


def get_accuracy(df):
    """ Input the directory of question file, and the corresponding difficulty, update the accuracy dictionary"""
    score_bleu = 0
    max_bleu = 0
    min_bleu = 0 

    for q, a in zip(df.en, df.fr):
        score = getscoreBLEU(q,a)
        max_bleu = max(max_bleu,score)
        min_bleu = min(min_bleu,score) 
        score_bleu += score
    score_bleu = score_bleu / df.shape[0]

    return(score_bleu,max_bleu,min_bleu,df.shape[0])
    
        


# Evaluate all files wait check that im loading the test data vs train data TODO
df = pd.read_csv("data/en-fr/en-fr-dataset/en-fr-test_dataframe.csv")
(score_bleu,max_bleu,min_bleu,tests) = get_accuracy(df)
print("----------------------------")
print(f"There are {tests} translations in total.\n\
       Using the BLEU metric The GPT-2 model had an average of {score_bleu}, \n\
       a max of {max_bleu} and a min of {min_bleu}. \n")