text-generation-webui/modules/logits.py

import torch

from modules import sampler_hijack, shared
from modules.exllama import ExllamaModel
from modules.exllamav2 import Exllamav2Model
from modules.llamacpp_model import LlamaCppModel
from modules.logging_colors import logger
from modules.text_generation import generate_reply

global_scores = None


def get_next_logits(prompt, state, use_samplers, previous):
    if shared.model is None:
        logger.error("No model is loaded! Select one in the Model tab.")
        return 'Error: No model is loaded1 Select one in the Model tab.', previous

    is_non_hf_exllamav2 = isinstance(shared.model, Exllamav2Model)
    is_non_hf_exllamav1 = isinstance(shared.model, ExllamaModel)
    is_non_hf_llamacpp = isinstance(shared.model, LlamaCppModel)

    if use_samplers:
        if any([is_non_hf_exllamav2, is_non_hf_exllamav1, is_non_hf_llamacpp]):
            logger.error("Sampler hijacking is not supported non-Huggingface loaders.")
            # sampling is all done in c for exllama, so it is really hard to hijack
            # it should be possible to hijack llamacpp sampler by hijacking all their sampling methods,
            # but it is not implemented yet
            return 'Error: Sampler hijacking is not supported non-Huggingface loaders. Please disable the "Use samplers" option.', previous

        state['max_new_tokens'] = 1
        state['auto_max_new_tokens'] = False
        for _ in generate_reply(prompt, state):
            pass

        scores = sampler_hijack.global_scores[-1]
    else:
        if is_non_hf_exllamav2 or is_non_hf_exllamav1:
            tokens = shared.tokenizer.encode(prompt).cuda()
            scores = shared.model.get_logits(tokens)[-1][-1]
        elif is_non_hf_llamacpp:
            tokens = shared.tokenizer.encode(prompt)
            scores = shared.model.get_logits(tokens)[-1][-1]
        else:
            tokens = shared.tokenizer.encode(prompt, return_tensors='pt').cuda()
            output = shared.model(input_ids=tokens)
            scores = output['logits'][-1][-1]

    probs = torch.softmax(scores, dim=-1, dtype=torch.float)
    topk_values, topk_indices = torch.topk(probs, k=25, largest=True, sorted=True)
    topk_values = [f"{float(i):.5f}" for i in topk_values]
    if is_non_hf_exllamav1 or is_non_hf_llamacpp:
        topk_indices = [i.expand((1, 1)) for i in topk_indices]

    tokens = [shared.tokenizer.decode(i) for i in topk_indices]
    if is_non_hf_llamacpp:
        tokens = [i.decode('utf-8') for i in tokens]  # llamacpp returns bytes, not str

    output = ''
    for row in list(zip(topk_values, tokens)):
        output += f"{row[0]}  -  {repr(row[1])[1:-1]}\n"

    return output, previous