mirror of
https://github.com/huggingface/text-generation-inference.git
synced 2025-04-20 06:12:07 +00:00
04e1af94d7
* feat: first draft load multiple lora * feat: load weights within layer and refactor lora pass * fix: refactor and reduce lora math * feat: baseline impl single request multi lora support * feat: prefer lorax implementation and port loading logic * fix: prefer adapter_data and refactors * feat: perfer loraxs custom punica kernels and add mlp loras * fix: adjust batch for bgmv * fix: adjust adapter_segments logic when in batch * fix: refactor and move changes to v3 proto * fix: pass model_id for all flash causal lms * fix: pass model_id for all causal and seq2seq lms * fix: add model_id to model test * feat: add lora support to mistral and refactors * feat: prefer model id in request * fix: include rust code for adapter id * feat: bump launcher and add new lora docs * feat: support base model generation and refactors * fix: rename doc to retry ci build * feat: support if vlm models * fix: add adapter_data param and avoid missing layers * fix: add adapter_data param to phi and neox * fix: update all models forwards to include adapter_data * fix: add model_id to IdeficsCausalLM * Update lora.md Fixed a typo * Update lora.md Fixing spam image * fix: add lora kernel to dockerfile, support running without kernels and refactors * fix: avoid dockerfile conflict * fix: refactors and adjust flash llama lora logic * fix: skip llama test due to CI issue (temp) * fix: skip llama test CI (temp) 2 * fix: revert skips and prefer updated ci token for tests * fix: refactors and helpful comments * fix: add noop in TensorParallelAdapterRowLinear too * fix: refactor and move shard_lora_weights logic * fix: exit early if no adapter_data --------- Co-authored-by: Derek <datavistics@gmail.com>
85 lines
2.8 KiB
Python
85 lines
2.8 KiB
Python
import torch
|
|
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM
|
|
from typing import List, Optional, Tuple
|
|
|
|
from text_generation_server.models import CausalLM
|
|
|
|
|
|
class RW(CausalLM):
|
|
def __init__(
|
|
self,
|
|
model_id: str,
|
|
revision: Optional[str] = None,
|
|
quantize: Optional[str] = None,
|
|
speculator: Optional[str] = None,
|
|
dtype: Optional[torch.dtype] = None,
|
|
trust_remote_code: bool = False,
|
|
):
|
|
if speculator:
|
|
raise RuntimeError("Medusa decoding is not enabled for AutoModel")
|
|
|
|
if torch.cuda.is_available():
|
|
device = torch.device("cuda")
|
|
dtype = torch.float16 if dtype is None else dtype
|
|
else:
|
|
if quantize:
|
|
raise ValueError("quantization is not available on CPU")
|
|
|
|
device = torch.device("cpu")
|
|
dtype = torch.float32 if dtype is None else dtype
|
|
|
|
tokenizer = AutoTokenizer.from_pretrained(
|
|
model_id,
|
|
revision=revision,
|
|
padding_side="left",
|
|
truncation_side="left",
|
|
trust_remote_code=trust_remote_code,
|
|
)
|
|
model = AutoModelForCausalLM.from_pretrained(
|
|
model_id,
|
|
revision=revision,
|
|
torch_dtype=dtype,
|
|
device_map=(
|
|
"auto"
|
|
if torch.cuda.is_available() and torch.cuda.device_count() > 1
|
|
else None
|
|
),
|
|
load_in_8bit=quantize == "bitsandbytes",
|
|
trust_remote_code=trust_remote_code,
|
|
)
|
|
if torch.cuda.is_available() and torch.cuda.device_count() == 1:
|
|
model = model.cuda()
|
|
|
|
if tokenizer.pad_token_id is None:
|
|
if model.config.pad_token_id is not None:
|
|
tokenizer.pad_token_id = model.config.pad_token_id
|
|
elif model.config.eos_token_id is not None:
|
|
tokenizer.pad_token_id = model.config.eos_token_id
|
|
elif tokenizer.eos_token_id is not None:
|
|
tokenizer.pad_token_id = tokenizer.eos_token_id
|
|
else:
|
|
tokenizer.add_special_tokens({"pad_token": "[PAD]"})
|
|
|
|
super(CausalLM, self).__init__(
|
|
model_id=model_id,
|
|
model=model,
|
|
tokenizer=tokenizer,
|
|
requires_padding=True,
|
|
dtype=dtype,
|
|
device=device,
|
|
)
|
|
|
|
def forward(
|
|
self, input_ids, attention_mask, position_ids, past_key_values: Optional = None
|
|
):
|
|
# Model Forward
|
|
outputs, speculative_logits = self.model.forward(
|
|
input_ids=input_ids,
|
|
attention_mask=attention_mask,
|
|
past_key_values=past_key_values,
|
|
use_cache=True,
|
|
)
|
|
|
|
return outputs.logits, speculative_logits, outputs.past_key_values
|