ceos/unsloth
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Update fp8.py

Browse source
This commit is contained in:
Daniel Han 2025-10-26 23:29:14 -07:00
commit 2c47b8a7ac
1 changed files with 4 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

8
unsloth/kernels/fp8.py
View file

@ -25,19 +25,19 @@ try:
from transformers.integrations.finegrained_fp8 import FP8Linear
except:
FP8Linear = None
logger.log("Unsloth: FP8 models need importing FP8Linear from `transformers.integrations.finegrained_fp8` but we don't see it.")
logger.info("Unsloth: FP8 models need importing FP8Linear from `transformers.integrations.finegrained_fp8` but we don't see it.")
try:
from transformers.integrations.fbgemm_fp8 import FbgemmFp8Linear
except:
FbgemmFp8Linear = None
logger.log("Unsloth: FP8 models need importing FbgemmFP8Linear from `transformers.integrations.fbgemm_fp8` but we don't see it.")
logger.info("Unsloth: FP8 models need importing FbgemmFP8Linear from `transformers.integrations.fbgemm_fp8` but we don't see it.")
try:
from fbgemm_gpu.experimental.gemm.triton_gemm.fp8_gemm import triton_quantize_fp8_block
except:
triton_quantize_fp8_block = None
logger.log("Unsloth: Could not find fbgemm_gpu.experimental.gemm.triton_gemm.fp8_gemm.triton_quantize_fp8_block")
logger.info("Unsloth: Could not find fbgemm_gpu.experimental.gemm.triton_gemm.fp8_gemm.triton_quantize_fp8_block")
try:
from torchao.prototype.blockwise_fp8_inference.blockwise_quantization import (
@ -45,7 +45,7 @@ try:
)
except:
torchao_blockwise_gemm = None
logger.log("Unsloth: Could not find torchao.prototype.blockwise_fp8_inference.blockwise_quantization.blockwise_fp8_gemm")
logger.info("Unsloth: Could not find torchao.prototype.blockwise_fp8_inference.blockwise_quantization.blockwise_fp8_gemm")
@triton.jit