Fixes to documentation

TRAINING.md

@@ -24,9 +24,9 @@ When doing human evaliuation the results for finetuned Catalan language model we
 
 Our hypothesis is that the evaluation on Common Voice gives better results because the model is overfitted and has lost generalization capabilities.
 
-**2. Model degrades according evaluation with other datasets**
+**3. Model degrades according evaluation with other datasets**
 
-Doing a more extensive evaluation shows:
+Results doing an evaluation with other datasets:
  
 |             | base        | sc-base     | small       | sc-small   |medium       | sc-medium   |
 | ----------- | ----------- | ----------- | ----------- |----------- | ----------- | ----------- |
@@ -35,7 +35,7 @@ Doing a more extensive evaluation shows:
 | Son_Goku_catalan_valencian_voice      | 51.90           | 85.44           | 39.87           |65.19           | 18.99| 71.52 
 | Universal_Declaration_of_Human_Rights      | 47.12           | 36.45           | 39.14          |75.59           | 44.37           | 27.79 
 
-As you can see, 
+As you can see, the fine-tunned models perform worse in most of the scenarios than OpenAI models.
  
 Legend:
 * "sc-" Indicates Softcatalà fine-tuned model
@@ -56,12 +56,11 @@ In our experiments
 | ----------- | ----------- |
 | OpenAI      | 27.32       |
 | Whisper.cpp 1.2.1   | 38.89 |
-| HuggingFace   | 93.54  |
+| HuggingFace 4.27.1  | 93.54  |
 | CTranslate2 3.10.3  | 43.68  |
 
 We strongly recommend using CTranslate2 as inference client.
 
-
 **5. Fine-tunning degrades timestamp prediction**
 
 Whisper uses timestamp tokens to indicate the timestamps of the transcribed texts.