Update README.md
9a1d0ed verified - assets Delete assets/5090.png
- 1.69 kB Upload ao-compile-ts results screenshot
- 3.4 kB Update README.md
- 335 MB Upload folder using huggingface_hub
- 1.59 kB Upload folder using huggingface_hub
- 29.2 GB Upload folder using huggingface_hub
- 243 Bytes Upload folder using huggingface_hub
- 663 Bytes Upload folder using huggingface_hub
- 1.67 MB Upload folder using huggingface_hub
- 123 kB Upload folder using huggingface_hub
model_fp8.pt Detected Pickle imports (17)
- "torchao.float8.inference.Float8MMConfig",
- "torch._utils._rebuild_tensor_v2",
- "collections.OrderedDict",
- "torch.serialization._get_layout",
- "torch.bfloat16",
- "torch._utils._rebuild_tensor_v3",
- "torch._utils._rebuild_wrapper_subclass",
- "torchao.quantization.Float8Tensor",
- "torch.BFloat16Storage",
- "torch.storage.UntypedStorage",
- "torch._tensor._rebuild_from_type_v2",
- "torch.float8_e4m3fn",
- "torch.device",
- "torchao.quantization.quantize_.workflows.float8.float8_tensor.QuantizeTensorToFloat8Kwargs",
- "torchao.quantization.granularity.PerTensor",
- "torch.FloatStorage",
- "torchao.quantization.quantize_.common.kernel_preference.KernelPreference"
How to fix it?
15.2 GB Rename bagel_fp8_quantized.pt to model_fp8.pt - 15.2 GB Upload model_fp8.safetensors with huggingface_hub
- 131 kB Create model_precision_report.txt
- 218 kB Upload model_safetensors_report.txt with huggingface_hub
- 7.03 MB Upload folder using huggingface_hub
- 7.31 kB Upload folder using huggingface_hub
- 205 Bytes Upload folder using huggingface_hub
- 2.78 MB Upload folder using huggingface_hub
