# Further Train a fine-tuned Modell?

**URL:** </static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Ffurther-train-a-fine-tuned-modell%2F5935%3E%5C
**Category:** Models\
**Created:** [May 2, 2021, 7:18pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Ffurther-train-a-fine-tuned-modell%2F5935 "2021-05-02T19:18:05Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![unknownTransformer](https://avatars.discourse-cdn.com/v4/letter/u/9f8e36/32.png) [@unknownTransformer](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FunknownTransformer)%5C
**Post date:** [May 2, 2021, 7:18pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Ffurther-train-a-fine-tuned-modell%2F5935%2F1 "2021-05-02T19:18:05Z")

</div>

Hi Guys,

how would i fruther train a already fine-tuned modell on new data?  
Iwant to (further) train a wav2vec2 Modell which was trained on CommonVoice on a new Dataset. Whould i have to train the modell from Scratch and use both datasets or could i just “finetune” the already fine tuned modell on new data? I tried multiple Learning-Rates and i just destroy the Modell (accuracy-wise) or the values dont change at all if i further train the modell

---

<div class="post-metadata">

**Author:** ![noskid](https://avatars.discourse-cdn.com/v4/letter/n/e19b73/32.png) [@noskid](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fnoskid)%5C
**Post date:** [June 10, 2021, 7:32am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Ffurther-train-a-fine-tuned-modell%2F5935%2F2 "2021-06-10T07:32:56Z")

</div>

Has anyone got any information on this?  
I am trying to train a model for a low resource language and I have a continuous flow of data instead of a collection. So, I want to find a way to continuously train my model further fine-tuned on OpenSLR data. Thank you!