giuliacassara

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

from torch.fx.experimental.proxy_tensor import make_fx
from torch._decomp import get_decompositions
import tempfile
import torch_mlir


def prepare_sentence_tokens(hf_model: str, sentence: str):

giuliacassara

module attributes {torch.debug_module_name = "GraphModule"} {
  func private @__torch__.torch.fx.graph_module.___torch_mangle_0.GraphModule.forward(%arg0: !torch.nn.Module<"__torch__.torch.fx.graph_module.___torch_mangle_0.GraphModule">, %arg1: !torch.tensor {torch.type_bound = !torch.vtensor<[50257,768],f32>}, %arg2: !torch.tensor {torch.type_bound = !torch.vtensor<[2304],f32>}, %arg3: !torch.tensor {torch.type_bound = !torch.vtensor<[768,2304],f32>}, %arg4: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg5: !torch.tensor {torch.type_bound = !torch.vtensor<[768,768],f32>}, %arg6: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg7: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg8: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg9: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg10: !torch.tensor {torch.type_bound = !torch.vtensor<[3072],f32>}, %arg11: !torch.tensor {torch.type_bound = !torch.vtensor<[768,3072],f32>}, %ar

giuliacassara

Building TensorFlow from source (TF 2.3.0, Ubuntu 20.04)

Why build from source?

The official instructions on installing TensorFlow are here: https://www.tensorflow.org/install. If you want to install TensorFlow just using pip, you are running a supported Ubuntu LTS distribution, and you're happy to install the respective tested CUDA versions (which often are outdated), by all means go ahead. A good alternative may be to run a Docker image.

I am usually unhappy with installing what in effect are pre-built binaries. These binaries are often not compatible with the Ubuntu version I am running, the CUDA version that I have installed, and so on. Furthermore, they may be slower than binaries optimized for the target architecture, since certain instructions are not being used (e.g. AVX2, FMA).

So installing TensorFlow from source becomes a necessity. The official instructions on building TensorFlow from source are here: ht

giuliacassara

import streamlit as st
import spacy
from spacy import displacy
import pandas as pd

from scispacy.umls_linking import UmlsEntityLinker
from scispacy.abbreviation import AbbreviationDetector


SPACY_MODEL_NAMES = ["en_core_sci_sm", "en_core_sci_md", "en_core_sci_lg"]