FiveTech Software tech support forums

Modesto Rocabado has shared with us this code to send messages and files using WhatsApp. We do appreciate your tests and feedback to fine tune it. Many thanks! FUNC SendToWhatsApp( cPhone, cMsg, aAttach )LOCAL oShell, aFiles := {}, aOthers  := {}, lRet If ...

Antonio Linares wrote:Dear Enrico,

I think it is a timing issue, maybe we invoke it too soon.

We need to find a way to fine tune it...

Yes, I was thinking along the same line. I tried with a call to SysWait() but to no avail. I have to try putting a delay in the C code.

Dear Enrico,

I think it is a timing issue, maybe we invoke it too soon.

We need to find a way to fine tune it...

... We have recently published a repo at https://github.com/FiveTechSoft/tinyMedical where you have the code to fine tune TinyLlama with your own data, in this case we have used a medical dataset. We encourage you to start building your own GGUFs. Full source code ...

hi, the "Problem" is to train (fine tune) a base model for own Data as it need much PC-Power Question : is it possible to "rent" PC-Power to train own Model :?: Dear Jimmy, You can use Google Colab with T4. You have a certain ...

hi,

the "Problem" is to train (fine tune) a base model for own Data as it need much PC-Power

Question : is it possible to "rent" PC-Power to train own Model

Dear Leandro, The first step is to create a dataset with questions and answers to train (fine tune) a base model such as Microsoft phi-2, TinyLlama, etc. Once trained, a GGUF file is generated, which can be used with FWH using the llama64.dll. This is the free and private ...

Estimado Leandro, Lo primero es crear un conjunto de datos (dataset) con preguntas y respuestas, con las que entrenar (fine tune) un modelo base como Microsoft phi-2, TinyLlama, etc Una vez entrenado, se genera un fichero GGUF que puede ya ser usado desde FWH usando llama64.dll Esta ...

https://www.kaggle.com/code/lucamassaron/fine-tune-phi-2-for-sentiment-analysis https://medium.com/@yernenip/optimizing-phi-2-a-deep-dive-into-fine-tuning-small-language-models-9d545ac90a99 https://medium.aiplanet.com/fine-tune-small-model-micphi-2-to-convert-natural-language-to-sql-32fc4f6ed40c

Locally using a fine tuned model with quantization: !pip install accelerate==0.25.0!pip install bitsandbytes==0.41.1!pip install datasets==2.14.6!pip install peft==0.6.2!pip install transformers==4.36.2!pip install torch==2.1.0!pip install einops==0.4.1  # Phi needs this one import torchfrom t...

Loading a fine tuned model from disk: (don't use this, it consumes a huge GPU memory! Use next post) !pip install accelerate==0.25.0!pip install bitsandbytes==0.41.1!pip install datasets==2.14.6!pip install peft==0.6.2!pip install transformers==4.36.2!pip install torch==2.1.0!pip install einops==0....

Locally saving a fine tuned model without modifying quantization: !pip install accelerate==0.25.0!pip install bitsandbytes==0.41.1!pip install datasets==2.14.6!pip install peft==0.6.2!pip install transformers==4.36.2!pip install torch==2.1.0!pip install einops==0.4.1  # Phi needs this one import to...

Locally saving a fine tuned model quantized: !pip install accelerate==0.25.0!pip install bitsandbytes==0.41.1!pip install datasets==2.14.6!pip install peft==0.6.2!pip install transformers==4.36.2!pip install torch==2.1.0!pip install einops==0.4.1  # Phi needs this one import torchfrom transformers ...

loading an extra trained layer to the base model Phi-2, based on: https://medium.com/@nimritakoul01/finetuning-microsoft-phi-2-small-language-model-on-veggo-dataset-using-qlora-8bcf70ab625e !pip install accelerate==0.25.0!pip install bitsandbytes==0.41.1!pip install datasets==2.14.6!pip inst...