royleibov commited on
Commit
8a3b670
1 Parent(s): ff9ff99

Update readme

Browse files
Files changed (1) hide show
  1. README.md +47 -0
README.md CHANGED
@@ -59,7 +59,54 @@ model-index:
59
  - name: Dev CER (+LM)
60
  type: cer
61
  value: 11.01
 
 
62
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
63
 
64
  # Fine-tuned XLSR-53 large model for speech recognition in English
65
 
 
59
  - name: Dev CER (+LM)
60
  type: cer
61
  value: 11.01
62
+ base_model:
63
+ - jonatasgrosman/wav2vec2-large-xlsr-53-english
64
  ---
65
+ # Disclaimer and Requirements
66
+
67
+ This model is a clone of [**jonatasgrosman/wav2vec2-large-xlsr-53-english**](https://huggingface.co/jonatasgrosman/wav2vec2-large-xlsr-53-english) compressed using ZipNN. Compressed losslessly to 88% its original size, ZipNN saved ~0.2GB in storage and potentially ~4PB in data transfer **monthly**.
68
+
69
+ ### Requirement
70
+
71
+ In order to use the model, ZipNN is necessary:
72
+ ```bash
73
+ pip install zipnn
74
+ ```
75
+ ### Use This Model
76
+ ```python
77
+ # Use a pipeline as a high-level helper
78
+ from transformers import pipeline
79
+ from zipnn import zipnn_hf
80
+
81
+ zipnn_hf()
82
+
83
+
84
+ pipe = pipeline("automatic-speech-recognition", model="royleibov/wav2vec2-large-xlsr-53-english")
85
+ ```
86
+ ```python
87
+ # Load model directly
88
+ from transformers import AutoProcessor, AutoModelForCTC
89
+ from zipnn import zipnn_hf
90
+
91
+ zipnn_hf()
92
+
93
+ processor = AutoProcessor.from_pretrained("royleibov/wav2vec2-large-xlsr-53-english")
94
+ model = AutoModelForCTC.from_pretrained("royleibov/wav2vec2-large-xlsr-53-english")
95
+ ```
96
+ ### ZipNN
97
+ ZipNN also allows you to seemlessly save local disk space in your cache after the model is downloaded.
98
+
99
+ To compress the cached model, simply run:
100
+ ```bash
101
+ python zipnn_compress_path.py safetensors --model royleibov/granite-3.0-8b-instruct-ZipNN-Compressed --hf_cache
102
+ ```
103
+
104
+ The model will be decompressed automatically and safely as long as `zipnn_hf()` is added at the top of the file like in the [example above](#use-this-model).
105
+
106
+ To decompress manualy, simply run:
107
+ ```bash
108
+ python zipnn_decompress_path.py --model royleibov/granite-3.0-8b-instruct-ZipNN-Compressed --hf_cache
109
+ ```
110
 
111
  # Fine-tuned XLSR-53 large model for speech recognition in English
112