File size: 2,766 Bytes
4923bc3 53eb687 4923bc3 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 | import torch
from transformers import PreTrainedModel, PretrainedConfig, GPT2TokenizerFast, \
AutoConfig, AutoModelForCausalLM
from transformers.modeling_outputs import CausalLMOutput
from model import FriendsTransformer
class PerkLMConfig(PretrainedConfig):
model_type = "perklm"
def __init__(self, d_model = 512, n_heads = 8, n_layers = 6,
d_ff = 2048, maxt = 512, dropout = 0.1, tokenizer_path = None, **kwargs):
super().__init__(**kwargs)
self.d_model = d_model
self.n_heads = n_heads
self.n_layers = n_layers
self.d_ff = d_ff
self.maxt = maxt
self.dropout = dropout
self.tokenizer_path = tokenizer_path
class PerkLM(PreTrainedModel):
config_class = PerkLMConfig
_tied_weights_keys = ["transformer.lm_head.weight"]
def __init__(self, config):
super().__init__(config)
tokenizer = GPT2TokenizerFast.from_pretrained(config.tokenizer_path)
self.transformer = FriendsTransformer(d_model = config.d_model,
n_heads = config.n_heads,
n_layers = config.n_layers,
d_ff = config.d_ff,
dropout = config.dropout,
maxt = config.maxt,
tokenizer = tokenizer)
self.post_init()
def forward(self, input_ids, attention_mask = None,
responder = None, labels = None, **kwargs):
batch = {'input_ids': input_ids,
'attention_mask': attention_mask if attention_mask is not None \
else torch.ones_like(input_ids),
'responder': responder}
logits = self.transformer(batch)
loss = None
if labels is not None:
loss = torch.nn.functional.cross_entropy(logits[:, :-1].reshape(-1, logits.size(-1)),
labels[:, 1:].reshape(-1), ignore_index = -100)
return CausalLMOutput(loss = loss, logits = logits)
def _save_pretrained_hook(self, *args, **kwargs):
pass
def state_dict(self, *args, **kwargs):
sd = super().state_dict(*args, **kwargs)
# strip complex64 RoPE buffers — recomputed on __init__
return {k: v for k, v in sd.items() if v.dtype != torch.complex64}
def tie_weights(self):
self.transformer.lm_head.weight = self.transformer.embedder.embedding.weight
AutoConfig.register("perklm", PerkLMConfig)
AutoModelForCausalLM.register(PerkLMConfig, PerkLM)
|