Dependency conflict: Cannot install TensorFlow in TCR-BERT’s conda env (Python 3.9, PyTorch 1.8.0)

[stellaxu2077]

Summary

I’m trying to install a version of TensorFlow that is compatible with the Python 3.9 + PyTorch 1.8.0 environment specified by TCR-BERT’s environment.yml. However, conda keeps reporting version conflicts, preventing me from installing TensorFlow.

---

Steps to Reproduce

1. Clone or download the TCR-BERT repository (if necessary).
2. Create and activate the TCR-BERT environment:

   conda env create -f environment.yml
   conda activate tcrbert```

3. Attempt to install TensorFlow:
   conda install tensorflow
4. Observe the dependency conflict error.

---

Actual Behavior & Full Error Log

Conda fails to resolve the dependencies and outputs the following conflict message:

LibMambaUnsatisfiableError: Encountered problems while solving:
  - nothing provides bleach 1.5.0 needed by tensorboard-1.6.0-py27hf484d3e_0

Could not solve for environment specs
The following packages are incompatible
├─ louvain =0.7.0 * is installable and it requires
│  └─ python-igraph =* *, which requires
│     └─ libblas >=3.8.0,<4.0a0 *, which requires
│        └─ blas =* mkl, which can be installed;
├─ pin on python 3.9.* =* * is installable and it requires
│  └─ python =3.9 *, which can be installed;
├─ pytorch =1.8.0 * is installable and it requires
│  └─ cudatoolkit >=10.2,<10.3 *, which can be installed;
├─ scipy =1.6.0 * is installable with the potential options
│  ├─ scipy [1.10.0|1.10.1|...|1.9.3] would require
│  │  └─ python >=3.8,<3.9.0a0 *, which conflicts with any installable versions previously reported;
│  ├─ scipy [1.10.0|1.10.1|...|1.9.3] would require
│  │  └─ blas =* openblas, which conflicts with any installable versions previously reported;
│  ├─ scipy [1.6.0|1.6.1|1.6.2|1.7.1|1.7.3] would require
│  │  └─ python >=3.7,<3.8.0a0 *, which conflicts with any installable versions previously reported;
│  └─ scipy 1.6.0, which can be installed;
├─ tensorboard =2.4.1 * is requested and can be installed;
└─ tensorflow =* * is not installable because there are no viable options
   ├─ tensorflow 2.4.1 would require
   │  └─ tensorflow-base ==2.4.1 gpu_py39h29c2da4_0, which requires
   │     └─ cudatoolkit >=10.1.243,<10.2.0a0 *, which conflicts with any installable versions previously reported;
   ├─ tensorflow [1.10.0|1.11.0|...|2.1.0] would require
   │  └─ python =2.7 *, which conflicts with any installable versions previously reported;
   ├─ tensorflow [1.10.0|1.9.0] would require
   │  └─ python =3.5 *, which conflicts with any installable versions previously reported;
   ├─ tensorflow [1.10.0|1.11.0|...|2.2.0] would require
   │  └─ python =3.6 *, which conflicts with any installable versions previously reported;
   ├─ tensorflow [1.10.0|1.8.0|1.9.0] would require
   │  └─ tensorflow-base [==1.10.0 gpu_py27h3435052_0|==1.10.0 gpu_py35h3435052_0|...|==1.9.0 gpu_py36h9f529ab_1], which requires
   │     └─ cudatoolkit =8.0 *, which does not exist (perhaps a missing channel);
   ├─ tensorflow [1.13.1|1.14.0|...|2.9.1] would require
   │  └─ python =3.7 *, which conflicts with any installable versions previously reported;
   ├─ tensorflow [2.10.0|2.11.0|...|2.9.1] would require
   │  └─ python [=3.10 *|>=3.10,<3.11.0a0 *], which conflicts with any installable versions previously reported;
   ├─ tensorflow [2.10.0|2.11.0|...|2.9.1] would require
   │  └─ python =3.8 *, which conflicts with any installable versions previously reported;
   ├─ tensorflow 2.10.0 would require
   │  └─ tensorboard =2.10 *, which conflicts with any installable versions previously reported;
   ├─ tensorflow 2.11.0 would require
   │  └─ tensorboard =2.11 *, which conflicts with any installable versions previously reported;
   ├─ tensorflow [2.12.0|2.17.0] would require
   │  └─ python [=3.11 *|>=3.11,<3.12.0a0 *], which conflicts with any installable versions previously reported;
   ├─ tensorflow 2.12.0 would require
   │  └─ tensorboard =2.12 *, which conflicts with any installable versions previously reported;
   ├─ tensorflow 2.17.0 would require
   │  └─ python >=3.12,<3.13.0a0 *, which conflicts with any installable versions previously reported;
   ├─ tensorflow 2.17.0 would require
   │  └─ tensorflow-base ==2.17.0 cpu_py39h6deec89_0, which requires
   │     └─ tensorboard >=2.17,<2.18 *, which conflicts with any installable versions previously reported;
   ├─ tensorflow 2.17.0 would require
   │  └─ __cuda =* *, which is missing on the system;
   ├─ tensorflow 2.4.1 would require
   │  └─ tensorflow-base [==2.4.1 eigen_py39h17880bf_0|==2.4.1 mkl_py39h43e0292_0], which requires
   │     └─ scipy >=1.6.1 * but there are no viable options
   │        ├─ scipy [1.10.0|1.10.1|...|1.9.3] would require
   │        │  └─ python >=3.10,<3.11.0a0 *, which conflicts with any installable versions previously reported;
   │        ├─ ...
   ├─ tensorflow 2.5.0 would require
   │  └─ tensorboard >=2.5.0 *, which conflicts ...
   ├─ tensorflow 2.6.0 would require
   │  └─ tensorboard >=2.6.0 *, which conflicts ...
   ├─ ...
Pins seem to be involved in the conflict. Currently pinned specs:
 - python=3.9

---

Additional Info

I have tried multiple times to adjust Python and other package versions within the TCR-BERT environment to avoid these conflicts, but I keep running into the same compatibility issues. The environment specifically pins Python at 3.9 and uses PyTorch 1.8.0 from the TCR-BERT environment.yml, and I need a TensorFlow version that won’t break these pins.
Any help or suggestions on how to resolve this dependency conflict and successfully install TensorFlow for training my model in this environment would be greatly appreciated. Thank you!