# gemma2    5.4GB
# llama3.3  43GB
# llama3.2  2.0GB
# gpt-oss:20b  14GB
# gemma3:4b     3.3GB
# deepseek-r1:8b   5.2GB

#  gpt-oss:20b  14GB
ollama pull gpt-oss:20b

# gemma3:4b     3.3GB
ollama pull gemma3:4b

#  deepseek-r1:8b   5.2GB
ollama pull deepseek-r1:8b



Training vs using Pre-trained Models

Using a pre-trained model involves either performing a forward pass for inference only or performing a full training cycle (forward pass, backpropagation, and weight updates) for fine-tuning. 
For inference, the pre-trained model's weights are fixed, and data is simply passed forward to generate predictions. For fine-tuning, you repeat the forward pass, calculate the error, and then use backpropagation to adjust the weights to better suit your specific dataset, starting with the pre-trained weights. 

LLM inferencing is the process of using a trained language model to generate a response to a user's prompt. It involves the model taking input text, processing it through complex mathematical operations to understand the prompt, and then generating an output token by token, based on the patterns it learned during training

# List llm models supported.
curl https://llm.ecloud.e3labs.net/api/tags | jq