Add evaluation details and env specs

README.md

@@ -13,16 +13,17 @@ base_model:
 
 # Wav2Vec2Bert Audio frame classifier for prosodic unit detection
 
-This model predicts prosodic units on speech.
-For each 20ms frame the model predicts 1 or 0, indicating whether there is a prosodic unit in 
-this frame or not.
+This model predicts prosodic units on speech. For each 20ms frame the model
+predicts 1 or 0, indicating whether there is a prosodic unit in this frame or
+not.
 
-This frame-level output can be grouped into events with the frames_to_intervals function provided in the 
-code snippets below.
+This frame-level output can be grouped into events with the frames_to_intervals
+function provided in the code snippets below.
 
-It is known that the model is unreliable if the audio starts or ends within a prosodic unit. This can be somewhat 
-circumvented by 1) using the largest possible chunks that will fit your machine and 2) use overlapping chunks
-and combining results smartly.
+It is known that the model is unreliable if the audio starts or ends within a
+prosodic unit. This can be somewhat circumvented by 1) using the largest
+possible chunks that will fit your machine and 2) use overlapping chunks and
+combining results smartly.
 
 
 
@@ -31,16 +32,39 @@ and combining results smartly.
 
 ### Model Description
 
-<!-- Provide a longer summary of what this model is. -->
 
-This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
 
-- **Developed by:** Peter Rupnik, Nikola Ljubešić, Darinka Verdonik, Simona Majheničy
+- **Developed by:** Peter Rupnik, Nikola Ljubešić, Darinka Verdonik, Simona
+  Majhenič
 - **Funded by:** MEZZANINE project
 - **Model type:** Wav2Vec2Bert for Audio Frame Classification
-- **Language(s) (NLP):** Trained and tested on Slovenian, ATM unclear if usable cross-lingually
+- **Language(s) (NLP):** Trained and tested on Slovenian, ATM unclear if usable
+  cross-lingually
 - **Finetuned from model:** facebook/w2v-bert-2.0
 
+The model was trained on [ROG-Art dataset](http://hdl.handle.net/11356/1992), on
+train split only.
+
+### Model performance
+
+We evaluate the model indirectly, and only care about the positive class:
+
+1. first prosodic units (intervals with start and end times, e.g. `[0.123,
+   5.546]`) are extracted from data and model outputs
+2. if a predicted prosodic unit has an overlapping counterpart in true prosodic
+   units, we count it as a True Positive. If there is no overlapping true
+   counterpart, we count it as a False Positive, and if we have a true prosodic
+   unit without a counterpart in predictions, we count that as a False Negative.
+3. Based on the TP, FN, FP numbers recall, precision, and F1 score is
+   calculated.
+
+In this fashion we obtain the following metrics:
+
+* Precision: 0.9423
+* Recall: 0.7802
+* F_1 score: 0.8538
+
+
 
 ## Uses
 
@@ -107,9 +131,9 @@ def evaluator(chunks):
         "prosodic_units": prosodic_units,
     }
 
-
-ds = Dataset.from_dict({"audio": [f, f]}).cast_column("audio", Audio(16000, mono=True))
-ds = ds.map(evaluator, batched=True, batch_size=2)
+# Create a dataset with a single instance and map our evaluator function on it:
+ds = Dataset.from_dict({"audio": [f]}).cast_column("audio", Audio(16000, mono=True))
+ds = ds.map(evaluator, batched=True, batch_size=1) # Adjust batch size according to your hardware specs
 print(ds["y_pred"][0])
 # Outputs: [0, 0, 1, 1, 1, 1, 1, ...]
 print(ds["y_pred_logits"][0])
@@ -124,9 +148,11 @@ print(ds["prosodic_units"][0])
 
 
 ### Inference on longer files
-If the file is too big for straight-forward inference, some chunking needs to be performed in order to process it. 
-We know that for starts and ends of chunks the probability of false negatives increases, so it is best to process the file
-with some overlap between chunks or split it on silence. We illustrate the former approach here:
+If the file is too big for straight-forward inference, some chunking needs to be
+performed in order to process it. We know that for starts and ends of chunks the
+probability of false negatives increases, so it is best to process the file with
+some overlap between chunks or split it on silence. We illustrate the former
+approach here:
 ```python
 import numpy as np
 
@@ -266,14 +292,15 @@ print(final_intervals)
 
 ## Training Details
 
-|hyperparameter|value|
-|---|---|
-|learning rate|3e-5|
-|batch size|1|
-|gradient accumulation steps|16|
-|num train epochs|20|
-|weight decay|0.01|
-
-## Evaluation
-
-
+| hyperparameter              | value |
+| --------------------------- | ----- |
+| learning rate               | 3e-5  |
+| batch size                  | 1     |
+| gradient accumulation steps | 16    |
+| num train epochs            | 20    |
+| weight decay                | 0.01  |
+
+Software environment can be found in mamba/conda [environment export yml
+file](transformers_env.yml). To recreate the environment with conda/mamba, run
+`mamba create -f transformers_env.yml` (replace mamba with conda if you don't
+use mamba).

transformers_env.yml

@@ -0,0 +1,423 @@
+name: transformers
+channels:
+  - pytorch
+  - nvidia
+  - conda-forge
+dependencies:
+  - _libgcc_mutex=0.1=conda_forge
+  - _openmp_mutex=4.5=2_kmp_llvm
+  - accelerate=0.33.0=pyhd8ed1ab_0
+  - aiohappyeyeballs=2.4.0=pyhd8ed1ab_0
+  - aiohttp=3.10.5=py311h61187de_0
+  - aiosignal=1.3.1=pyhd8ed1ab_0
+  - anyio=4.4.0=pyhd8ed1ab_0
+  - aom=3.5.0=h27087fc_0
+  - argon2-cffi=23.1.0=pyhd8ed1ab_0
+  - argon2-cffi-bindings=21.2.0=py311h9ecbd09_5
+  - arrow=1.3.0=pyhd8ed1ab_0
+  - asttokens=2.4.1=pyhd8ed1ab_0
+  - async-lru=2.0.4=pyhd8ed1ab_0
+  - atk-1.0=2.38.0=h04ea711_2
+  - attrs=24.2.0=pyh71513ae_0
+  - audioread=3.0.1=py311h38be061_1
+  - aws-c-auth=0.7.22=h96bc93b_2
+  - aws-c-cal=0.6.14=h88a6e22_1
+  - aws-c-common=0.9.19=h4ab18f5_0
+  - aws-c-compression=0.2.18=h83b837d_6
+  - aws-c-event-stream=0.4.2=ha47c788_12
+  - aws-c-http=0.8.1=h29d6fba_17
+  - aws-c-io=0.14.8=h21d4f22_5
+  - aws-c-mqtt=0.10.4=h759edc4_4
+  - aws-c-s3=0.5.9=h594631b_3
+  - aws-c-sdkutils=0.1.16=h83b837d_2
+  - aws-checksums=0.1.18=h83b837d_6
+  - aws-crt-cpp=0.26.9=he3a8b3b_0
+  - aws-sdk-cpp=1.11.329=hba8bd5f_3
+  - babel=2.14.0=pyhd8ed1ab_0
+  - baumwelch=0.3.9=h434a139_3
+  - beautifulsoup4=4.12.3=pyha770c72_0
+  - biopython=1.79=py311hd4cff14_3
+  - blas=1.0=mkl
+  - bleach=6.1.0=pyhd8ed1ab_0
+  - brotli=1.1.0=hd590300_1
+  - brotli-bin=1.1.0=hd590300_1
+  - brotli-python=1.1.0=py311hb755f60_1
+  - bzip2=1.0.8=h4bc722e_7
+  - c-ares=1.33.0=ha66036c_0
+  - ca-certificates=2024.12.14=hbcca054_0
+  - cached-property=1.5.2=hd8ed1ab_1
+  - cached_property=1.5.2=pyha770c72_1
+  - cairo=1.18.0=hebfffa5_3
+  - certifi=2024.12.14=pyhd8ed1ab_0
+  - cffi=1.17.0=py311ha8e6434_0
+  - charset-normalizer=3.3.2=pyhd8ed1ab_0
+  - click=8.1.7=unix_pyh707e725_0
+  - cloudpickle=3.1.0=pyhd8ed1ab_2
+  - colorama=0.4.6=pyhd8ed1ab_0
+  - comm=0.2.2=pyhd8ed1ab_0
+  - contourpy=1.2.1=py311h9547e67_0
+  - cuda-cudart=11.8.89=0
+  - cuda-cupti=11.8.87=0
+  - cuda-libraries=11.8.0=0
+  - cuda-nvrtc=11.8.89=0
+  - cuda-nvtx=11.8.86=0
+  - cuda-runtime=11.8.0=0
+  - cuda-version=12.6=3
+  - cycler=0.12.1=pyhd8ed1ab_0
+  - cython=3.0.11=py311h55d416d_3
+  - dataclassy=1.0.1=pyhd8ed1ab_0
+  - datasets=2.21.0=pyhd8ed1ab_0
+  - debugpy=1.8.5=py311hfdbb021_1
+  - decorator=5.1.1=pyhd8ed1ab_0
+  - defusedxml=0.7.1=pyhd8ed1ab_0
+  - dill=0.3.8=pyhd8ed1ab_0
+  - entrypoints=0.4=pyhd8ed1ab_0
+  - exceptiongroup=1.2.2=pyhd8ed1ab_0
+  - executing=2.1.0=pyhd8ed1ab_0
+  - expat=2.6.2=h59595ed_0
+  - ffmpeg=5.1.2=gpl_h8dda1f0_106
+  - filelock=3.15.4=pyhd8ed1ab_0
+  - font-ttf-dejavu-sans-mono=2.37=hab24e00_0
+  - font-ttf-inconsolata=3.000=h77eed37_0
+  - font-ttf-source-code-pro=2.038=h77eed37_0
+  - font-ttf-ubuntu=0.83=h77eed37_2
+  - fontconfig=2.14.2=h14ed4e7_0
+  - fonts-conda-ecosystem=1=0
+  - fonts-conda-forge=1=0
+  - fonttools=4.53.1=py311h61187de_0
+  - fqdn=1.5.1=pyhd8ed1ab_0
+  - freetype=2.12.1=h267a509_2
+  - fribidi=1.0.10=h36c2ea0_0
+  - frozenlist=1.4.1=py311h459d7ec_0
+  - fsspec=2024.5.0=pyhff2d567_0
+  - gdk-pixbuf=2.42.12=hb9ae30d_0
+  - gettext=0.22.5=he02047a_3
+  - gettext-tools=0.22.5=he02047a_3
+  - gflags=2.2.2=he1b5a44_1004
+  - glog=0.7.1=hbabe93e_0
+  - gmp=6.3.0=hac33072_2
+  - gmpy2=2.1.5=py311hc4f1f91_1
+  - gnutls=3.7.9=hb077bed_0
+  - graphite2=1.3.13=h59595ed_1003
+  - graphviz=12.0.0=hba01fac_0
+  - greenlet=3.1.1=py311hfdbb021_0
+  - gtk2=2.24.33=h6470451_5
+  - gts=0.7.6=h977cf35_4
+  - h11=0.14.0=pyhd8ed1ab_0
+  - h2=4.1.0=pyhd8ed1ab_0
+  - harfbuzz=9.0.0=hda332d3_1
+  - hdbscan=0.8.38.post1=py311h9f3472d_2
+  - hpack=4.0.0=pyh9f0ad1d_0
+  - httpcore=1.0.5=pyhd8ed1ab_0
+  - httpx=0.27.2=pyhd8ed1ab_0
+  - huggingface_hub=0.24.6=pyhd8ed1ab_0
+  - hyperframe=6.0.1=pyhd8ed1ab_0
+  - icu=75.1=he02047a_0
+  - idna=3.7=pyhd8ed1ab_0
+  - importlib-metadata=8.4.0=pyha770c72_0
+  - importlib_metadata=8.4.0=hd8ed1ab_0
+  - importlib_resources=6.4.5=pyhd8ed1ab_0
+  - ipykernel=6.29.5=pyh3099207_0
+  - ipython=8.27.0=pyh707e725_0
+  - ipywidgets=8.1.5=pyhd8ed1ab_0
+  - isoduration=20.11.0=pyhd8ed1ab_0
+  - jedi=0.19.1=pyhd8ed1ab_0
+  - jinja2=3.1.4=pyhd8ed1ab_0
+  - joblib=1.4.2=pyhd8ed1ab_0
+  - json5=0.9.25=pyhd8ed1ab_0
+  - jsonpointer=3.0.0=py311h38be061_1
+  - jsonschema=4.23.0=pyhd8ed1ab_0
+  - jsonschema-specifications=2023.12.1=pyhd8ed1ab_0
+  - jsonschema-with-format-nongpl=4.23.0=hd8ed1ab_0
+  - jupyter=1.1.1=pyhd8ed1ab_0
+  - jupyter-lsp=2.2.5=pyhd8ed1ab_0
+  - jupyter_client=8.6.2=pyhd8ed1ab_0
+  - jupyter_console=6.6.3=pyhd8ed1ab_0
+  - jupyter_core=5.7.2=py311h38be061_0
+  - jupyter_events=0.10.0=pyhd8ed1ab_0
+  - jupyter_server=2.14.2=pyhd8ed1ab_0
+  - jupyter_server_terminals=0.5.3=pyhd8ed1ab_0
+  - jupyterlab=4.2.5=pyhd8ed1ab_0
+  - jupyterlab_pygments=0.3.0=pyhd8ed1ab_1
+  - jupyterlab_server=2.27.3=pyhd8ed1ab_0
+  - jupyterlab_widgets=3.0.13=pyhd8ed1ab_0
+  - kaldi=5.5.1112=cpu_hd7b63f8_3
+  - kalpy=0.6.7=py311hd18a35c_0
+  - keyutils=1.6.1=h166bdaf_0
+  - kiwisolver=1.4.5=py311h9547e67_1
+  - kneed=0.8.5=pyhd8ed1ab_0
+  - krb5=1.21.3=h659f571_0
+  - lame=3.100=h166bdaf_1003
+  - lazy-loader=0.4=pyhd8ed1ab_1
+  - lazy_loader=0.4=pyhd8ed1ab_1
+  - lcms2=2.16=hb7c19ff_0
+  - ld_impl_linux-64=2.40=hf3520f5_7
+  - lerc=4.0.0=h27087fc_0
+  - libabseil=20240116.2=cxx17_he02047a_1
+  - libarrow=16.1.0=hcb6531f_6_cpu
+  - libarrow-acero=16.1.0=hac33072_6_cpu
+  - libarrow-dataset=16.1.0=hac33072_6_cpu
+  - libarrow-substrait=16.1.0=h7e0c224_6_cpu
+  - libasprintf=0.22.5=he8f35ee_3
+  - libasprintf-devel=0.22.5=he8f35ee_3
+  - libblas=3.9.0=16_linux64_mkl
+  - libbrotlicommon=1.1.0=hd590300_1
+  - libbrotlidec=1.1.0=hd590300_1
+  - libbrotlienc=1.1.0=hd590300_1
+  - libcblas=3.9.0=16_linux64_mkl
+  - libcrc32c=1.1.2=h9c3ff4c_0
+  - libcublas=11.11.3.6=0
+  - libcufft=10.9.0.58=0
+  - libcufile=1.11.0.15=0
+  - libcurand=10.3.7.37=0
+  - libcurl=8.8.0=hca28451_1
+  - libcusolver=11.4.1.48=0
+  - libcusparse=11.7.5.86=0
+  - libdeflate=1.21=h4bc722e_0
+  - libdrm=2.4.123=hb9d3cd8_0
+  - libedit=3.1.20191231=he28a2e2_2
+  - libev=4.33=hd590300_2
+  - libevent=2.1.12=hf998b51_1
+  - libexpat=2.6.2=h59595ed_0
+  - libffi=3.4.2=h7f98852_5
+  - libflac=1.4.3=h59595ed_0
+  - libgcc=14.1.0=h77fa898_1
+  - libgcc-ng=14.1.0=h69a702a_1
+  - libgd=2.3.3=hd3e95f3_10
+  - libgettextpo=0.22.5=he02047a_3
+  - libgettextpo-devel=0.22.5=he02047a_3
+  - libgfortran=3.0.0=1
+  - libgfortran-ng=14.1.0=h69a702a_0
+  - libgfortran5=14.1.0=hc5f4f2c_0
+  - libglib=2.82.1=h2ff4ddf_0
+  - libgoogle-cloud=2.24.0=h2736e30_0
+  - libgoogle-cloud-storage=2.24.0=h3d9a0c8_0
+  - libgrpc=1.62.2=h15f2491_0
+  - libhwloc=2.11.1=default_hecaa2ac_1000
+  - libiconv=1.17=hd590300_2
+  - libidn2=2.3.7=hd590300_0
+  - libjpeg-turbo=3.0.0=hd590300_1
+  - liblapack=3.9.0=16_linux64_mkl
+  - liblapacke=3.9.0=16_linux64_mkl
+  - libllvm14=14.0.6=hcd5def8_4
+  - libnghttp2=1.58.0=h47da74e_1
+  - libnpp=11.8.0.86=0
+  - libnsl=2.0.1=hd590300_0
+  - libnvjpeg=11.9.0.86=0
+  - libogg=1.3.5=h4ab18f5_0
+  - libopus=1.3.1=h7f98852_1
+  - libparquet=16.1.0=h6a7eafb_6_cpu
+  - libpciaccess=0.18=hd590300_0
+  - libpng=1.6.43=h2797004_0
+  - libpq=16.4=h2d7952a_2
+  - libprotobuf=4.25.3=h08a7969_0
+  - libre2-11=2023.09.01=h5a48ba9_2
+  - librosa=0.10.2.post1=pyhd8ed1ab_0
+  - librsvg=2.58.4=hc0ffecb_0
+  - libsndfile=1.2.2=hc60ed4a_1
+  - libsodium=1.0.20=h4ab18f5_0
+  - libsqlite=3.46.0=hde9e2c9_0
+  - libssh2=1.11.0=h0841786_0
+  - libstdcxx=14.1.0=hc0a3c3a_1
+  - libstdcxx-ng=14.1.0=h4852527_1
+  - libtasn1=4.19.0=h166bdaf_0
+  - libthrift=0.19.0=hb90f79a_1
+  - libtiff=4.7.0=h6565414_0
+  - libunistring=0.9.10=h7f98852_0
+  - libutf8proc=2.8.0=h166bdaf_0
+  - libuuid=2.38.1=h0b41bf4_0
+  - libva=2.18.0=h0b41bf4_0
+  - libvorbis=1.3.7=h9c3ff4c_0
+  - libvpx=1.11.0=h9c3ff4c_3
+  - libwebp-base=1.4.0=hd590300_0
+  - libxcb=1.17.0=h8a09558_0
+  - libxcrypt=4.4.36=hd590300_1
+  - libxml2=2.12.7=he7c6b58_4
+  - libxslt=1.1.39=h76b75d6_0
+  - libzlib=1.3.1=h4ab18f5_1
+  - llvm-openmp=15.0.7=h0cdce71_0
+  - llvmlite=0.42.0=py311ha6695c7_1
+  - lxml=5.3.0=py311hcfaa980_1
+  - lz4-c=1.9.4=hcb278e6_0
+  - mad=0.15.1b=h9c3ff4c_1
+  - markdown-it-py=3.0.0=pyhd8ed1ab_0
+  - markupsafe=2.1.5=py311h459d7ec_0
+  - matplotlib-base=3.9.2=py311h74b4f7c_0
+  - matplotlib-inline=0.1.7=pyhd8ed1ab_0
+  - mdurl=0.1.2=pyhd8ed1ab_0
+  - mistune=3.0.2=pyhd8ed1ab_0
+  - mkl=2022.2.1=h84fe81f_16997
+  - montreal-forced-aligner=3.1.3=pyhd8ed1ab_0
+  - mpc=1.3.1=hfe3b2da_0
+  - mpfr=4.2.1=h38ae2d0_2
+  - mpg123=1.32.6=h59595ed_0
+  - mpmath=1.3.0=pyhd8ed1ab_0
+  - msgpack-python=1.0.8=py311h52f7536_0
+  - multidict=6.0.5=py311h459d7ec_0
+  - multiprocess=0.70.16=py311h459d7ec_0
+  - munkres=1.1.4=pyh9f0ad1d_0
+  - nbclient=0.10.0=pyhd8ed1ab_0
+  - nbconvert-core=7.16.4=pyhd8ed1ab_1
+  - nbformat=5.10.4=pyhd8ed1ab_0
+  - ncurses=6.5=h59595ed_0
+  - nest-asyncio=1.6.0=pyhd8ed1ab_0
+  - nettle=3.9.1=h7ab15ed_0
+  - networkx=3.3=pyhd8ed1ab_1
+  - ngram=1.3.16=h434a139_1
+  - notebook=7.2.2=pyhd8ed1ab_0
+  - notebook-shim=0.2.4=pyhd8ed1ab_0
+  - numba=0.59.1=py311h96b013e_0
+  - numpy=1.26.4=py311h64a7726_0
+  - openfst=1.8.3=h00ab1b0_2
+  - openh264=2.3.1=hcb278e6_2
+  - openjpeg=2.5.2=h488ebb8_0
+  - openssl=3.4.0=hb9d3cd8_0
+  - orc=2.0.1=h17fec99_1
+  - overrides=7.7.0=pyhd8ed1ab_0
+  - p11-kit=0.24.1=hc5aa10d_0
+  - packaging=24.1=pyhd8ed1ab_0
+  - pandas=2.2.2=py311h14de704_1
+  - pandocfilters=1.5.0=pyhd8ed1ab_0
+  - pango=1.54.0=h4c5309f_1
+  - parso=0.8.4=pyhd8ed1ab_0
+  - pcre2=10.44=hba22ea6_2
+  - pexpect=4.9.0=pyhd8ed1ab_0
+  - pgvector=0.7.4=h2f8f9d6_0
+  - pgvector-python=0.3.4=pyh1ff3077_0
+  - pickleshare=0.7.5=py_1003
+  - pillow=10.4.0=py311h4aec55e_1
+  - pip=24.2=pyhd8ed1ab_0
+  - pixman=0.43.2=h59595ed_0
+  - pkgutil-resolve-name=1.3.10=pyhd8ed1ab_1
+  - platformdirs=4.2.2=pyhd8ed1ab_0
+  - polars=1.17.1=py311h03f6b34_0
+  - pooch=1.8.2=pyhd8ed1ab_0
+  - postgresql=16.4=hb2eb5c0_2
+  - praatio=6.0.0=pyhd8ed1ab_0
+  - prometheus_client=0.20.0=pyhd8ed1ab_0
+  - prompt-toolkit=3.0.47=pyha770c72_0
+  - prompt_toolkit=3.0.47=hd8ed1ab_0
+  - psutil=6.0.0=py311h331c9d8_0
+  - psycopg2=2.9.9=py311h03dec38_0
+  - pthread-stubs=0.4=h36c2ea0_1001
+  - ptyprocess=0.7.0=pyhd3deb0d_0
+  - pure_eval=0.2.3=pyhd8ed1ab_0
+  - pyarrow=16.1.0=py311h781c19f_1
+  - pyarrow-core=16.1.0=py311h8e2c35d_1_cpu
+  - pyarrow-hotfix=0.6=pyhd8ed1ab_0
+  - pycparser=2.22=pyhd8ed1ab_0
+  - pydub=0.25.1=pyhd8ed1ab_0
+  - pygments=2.18.0=pyhd8ed1ab_0
+  - pynini=2.1.6=py311h9547e67_0
+  - pynvml=11.4.1=pyhd8ed1ab_0
+  - pyparsing=3.1.2=pyhd8ed1ab_0
+  - pysocks=1.7.1=pyha2e5f31_6
+  - pysoundfile=0.12.1=pyhd8ed1ab_0
+  - python=3.11.9=hb806964_0_cpython
+  - python-dateutil=2.9.0=pyhd8ed1ab_0
+  - python-fastjsonschema=2.20.0=pyhd8ed1ab_0
+  - python-json-logger=2.0.7=pyhd8ed1ab_0
+  - python-tzdata=2024.1=pyhd8ed1ab_0
+  - python-xxhash=3.5.0=py311h61187de_0
+  - python_abi=3.11=5_cp311
+  - pytorch=2.4.0=py3.11_cuda11.8_cudnn9.1.0_0
+  - pytorch-cuda=11.8=h7e8668a_5
+  - pytorch-mutex=1.0=cuda
+  - pytz=2024.1=pyhd8ed1ab_0
+  - pyyaml=6.0.2=py311h61187de_0
+  - pyzmq=26.2.0=py311h7deb3e3_2
+  - qhull=2020.2=h434a139_5
+  - re2=2023.09.01=h7f4b329_2
+  - readline=8.2=h8228510_1
+  - referencing=0.35.1=pyhd8ed1ab_0
+  - regex=2024.7.24=py311h61187de_0
+  - requests=2.32.3=pyhd8ed1ab_0
+  - rfc3339-validator=0.1.4=pyhd8ed1ab_0
+  - rfc3986-validator=0.1.1=pyh9f0ad1d_0
+  - rich=13.8.1=pyhd8ed1ab_0
+  - rich-click=1.8.3=pyhd8ed1ab_0
+  - rpds-py=0.20.0=py311h9e33e62_1
+  - s2n=1.4.15=he19d79f_0
+  - safetensors=0.4.4=py311hb3a8bbb_0
+  - scalene=1.5.41=py311h4332511_0
+  - scikit-learn=1.2.2=py311hc009520_2
+  - scipy=1.14.0=py311h0a5b728_2
+  - send2trash=1.8.3=pyh0d859eb_0
+  - setuptools=72.2.0=pyhd8ed1ab_0
+  - six=1.16.0=pyh6c4a22f_0
+  - snappy=1.2.1=ha2e4443_0
+  - sniffio=1.3.1=pyhd8ed1ab_0
+  - soupsieve=2.5=pyhd8ed1ab_1
+  - sox=14.4.2=h32e7c5b_1019
+  - soxr=0.1.3=h0b41bf4_3
+  - soxr-python=0.4.0=py311h07ce7c0_0
+  - sqlalchemy=2.0.35=py311h9ecbd09_0
+  - sqlite=3.46.0=h6d4b2fc_0
+  - stack_data=0.6.2=pyhd8ed1ab_0
+  - svt-av1=1.4.1=hcb278e6_0
+  - sympy=1.13.2=pypyh2585a3b_103
+  - tbb=2021.12.0=h434a139_3
+  - terminado=0.18.1=pyh0d859eb_0
+  - threadpoolctl=3.5.0=pyhc1e730c_0
+  - tinycss2=1.3.0=pyhd8ed1ab_0
+  - tk=8.6.13=noxft_h4845f30_101
+  - tokenizers=0.19.1=py311h6640629_0
+  - tomli=2.0.1=pyhd8ed1ab_0
+  - torchaudio=2.4.0=py311_cu118
+  - torchtriton=3.0.0=py311
+  - torchvision=0.19.0=py311_cu118
+  - tornado=6.4.1=py311h9ecbd09_1
+  - tqdm=4.66.5=pyhd8ed1ab_0
+  - traitlets=5.14.3=pyhd8ed1ab_0
+  - transformers=4.44.1=pyhd8ed1ab_0
+  - types-python-dateutil=2.9.0.20240906=pyhd8ed1ab_0
+  - typing-extensions=4.12.2=hd8ed1ab_0
+  - typing_extensions=4.12.2=pyha770c72_0
+  - typing_utils=0.1.0=pyhd8ed1ab_0
+  - tzcode=2024b=hb9d3cd8_0
+  - tzdata=2024a=h0c530f3_0
+  - uri-template=1.3.0=pyhd8ed1ab_0
+  - urllib3=2.2.2=pyhd8ed1ab_1
+  - wcwidth=0.2.13=pyhd8ed1ab_0
+  - webcolors=24.8.0=pyhd8ed1ab_0
+  - webencodings=0.5.1=pyhd8ed1ab_2
+  - websocket-client=1.8.0=pyhd8ed1ab_0
+  - wheel=0.44.0=pyhd8ed1ab_0
+  - widgetsnbextension=4.0.13=pyhd8ed1ab_0
+  - x264=1!164.3095=h166bdaf_2
+  - x265=3.5=h924138e_3
+  - xorg-fixesproto=5.0=h7f98852_1002
+  - xorg-kbproto=1.0.7=h7f98852_1002
+  - xorg-libice=1.1.1=hb9d3cd8_1
+  - xorg-libsm=1.2.4=he73a12e_1
+  - xorg-libx11=1.8.10=h4f16b4b_0
+  - xorg-libxau=1.0.11=hd590300_0
+  - xorg-libxdmcp=1.1.3=h7f98852_0
+  - xorg-libxext=1.3.4=h0b41bf4_2
+  - xorg-libxfixes=5.0.3=h7f98852_1004
+  - xorg-libxrender=0.9.11=hb9d3cd8_1
+  - xorg-xextproto=7.3.0=h0b41bf4_1003
+  - xorg-xorgproto=2024.1=hb9d3cd8_1
+  - xorg-xproto=7.0.31=h7f98852_1007
+  - xxhash=0.8.2=hd590300_0
+  - xz=5.2.6=h166bdaf_0
+  - yaml=0.2.5=h7f98852_2
+  - yarl=1.9.4=py311h459d7ec_0
+  - zeromq=4.3.5=ha4adb4c_5
+  - zip=3.0=hd590300_3
+  - zipp=3.20.0=pyhd8ed1ab_0
+  - zlib=1.3.1=h4ab18f5_1
+  - zstandard=0.23.0=py311h5cd10c7_0
+  - zstd=1.5.6=ha6fb4c9_0
+  - pip:
+      - annotated-types==0.7.0
+      - bokeh==3.6.2
+      - fastapi==0.112.1
+      - line-profiler==4.1.3
+      - pydantic==2.8.2
+      - pydantic-core==2.20.1
+      - starlette==0.38.2
+      - transliterate==1.10.2
+      - uvicorn==0.30.6
+      - xyzservices==2024.9.0
+prefix: /home/peterr/mambaforge/envs/transformers