|
|
--- |
|
|
title: "Training Space for Job 304268a6-2e7f-4d73-9cb6-e538064f8e80" |
|
|
emoji: "🚀" |
|
|
colorFrom: "blue" |
|
|
colorTo: "green" |
|
|
sdk: "docker" |
|
|
python_version: "3.10" |
|
|
suggested_hardware: "t4-small" |
|
|
app_file: "train_script_wrapper.py" |
|
|
pinned: false |
|
|
short_description: "pythia-70m-deduped HF-mb28snxb" |
|
|
models: |
|
|
- "EleutherAI/pythia-70m-deduped" |
|
|
datasets: |
|
|
- "Testys/dataset_for_job_304268a6_2e7f_4d73_9cb6_e538064f8e80" |
|
|
tags: |
|
|
- "automated-training" |
|
|
- "causal-lm" |
|
|
- "fine-tuning" |
|
|
- "lora" |
|
|
- "mlops" |
|
|
- "text-generation" |
|
|
preload_from_hub: |
|
|
- "EleutherAI/pythia-70m-deduped" |
|
|
--- |
|
|
|
|
|
# Training Space for MLOps Job `304268a6-2e7f-4d73-9cb6-e538064f8e80` |
|
|
|
|
|
This Space is automatically generated to run a LoRA (Low-Rank Adaptation) fine-tuning job. |
|
|
It utilizes a Docker environment to execute the `train_text_lora.py` script using parameters defined by the MLOps job configuration. |
|
|
|
|
|
## Job Overview |
|
|
- **Job ID:** `304268a6-2e7f-4d73-9cb6-e538064f8e80` |
|
|
- **Job Name:** `pythia-70m-deduped HF-mb28snxb` |
|
|
- **Model Type:** `CAUSAL_LM` |
|
|
- **Base Model (for fine-tuning):** `EleutherAI/pythia-70m-deduped` |
|
|
- **Dataset (on Hugging Face Hub):** `Testys/dataset_for_job_304268a6_2e7f_4d73_9cb6_e538064f8e80` |
|
|
|
|
|
## Execution Details |
|
|
The core training logic is encapsulated in `train_text_lora.py`, which is orchestrated by `train_script_wrapper.py` within this Space. |
|
|
Hyperparameters and script configurations are passed dynamically to the training script. |
|
|
|
|
|
## Outputs |
|
|
Outputs from the training process, such as the LoRA adapter and training metrics, will be pushed to the following Hugging Face Hub model repository upon successful completion: |
|
|
[Target repository to be configured](https://huggingface.co/#) |
|
|
|
|
|
## Monitoring |
|
|
Check the **Logs** tab of this Space for real-time training progress, standard output, and any error messages from the execution script. |
|
|
|
