huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-09-11 12:24:53 +00:00
Code Issues Packages Projects Releases Wiki Activity

use GPT2TokenizerFast by default

Browse Source
This commit is contained in:
OlivierDehaene 2024-03-29 18:46:28 +01:00
parent dcfefc425a
commit 275a61aae6
1 changed files with 22 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

33
server/text_generation_server/models/flash_dbrx.py
View File

@ -38,7 +38,7 @@ class FlashDbrx(FlashCausalLM):
raise NotImplementedError("FlashDBRX is only available on GPU")
try:
tokenizer = AutoTokenizer.from_pretrained(
tokenizer = GPT2TokenizerFast.from_pretrained(
model_id,
revision=revision,
padding_side="left",
@ -48,16 +48,27 @@ class FlashDbrx(FlashCausalLM):
from_slow=False,
)
except:
# FIXME: change back to model id once the tokenizer.json is merged
tokenizer = GPT2TokenizerFast.from_pretrained(
"Xenova/dbrx-instruct-tokenizer",
revision=revision,
padding_side="left",
truncation_side="left",
trust_remote_code=trust_remote_code,
use_fast=True,
from_slow=False,
)
try:
tokenizer = AutoTokenizer.from_pretrained(
model_id,
revision=revision,
padding_side="left",
truncation_side="left",
trust_remote_code=trust_remote_code,
use_fast=True,
from_slow=False,
)
except:
# FIXME: change back to model id once the tokenizer.json is merged
tokenizer = GPT2TokenizerFast.from_pretrained(
"Xenova/dbrx-instruct-tokenizer",
revision=revision,
padding_side="left",
truncation_side="left",
trust_remote_code=trust_remote_code,
use_fast=True,
from_slow=False,
)
config = DbrxConfig.from_pretrained(
model_id, revision=revision, trust_remote_code=trust_remote_code