This is a Riva-generated LoRA fine-tune of openai-community/gpt2-xl for classification, trained on lakshmi25npathi_imdb-dataset-of-50k-movie-reviews_review_sentiment_train.jsonl with 4,102 examples. Its final classification_f1 score was 0.8882.

Use Cases

Task-specific text generation, classification, extraction, or rewriting workflows.
Prototyping a model behavior before investing in larger training runs.
Continuing the same training setup with your own data in Riva.

Continue Training in Riva

This model was trained with Riva.

Start from this checkpoint, choose your own dataset, and create your own version in minutes:

https://app.riva.run/import?hf=HassanJbr/gpt2-xl-classification-lora-riva

Training Summary

Field	Value
Base model	`openai-community/gpt2-xl`
Method	LoRA
Checkpoint type	lora
Publish version	v2
GPU	T4
Max steps	50
Learning rate	0.0001
Seed	42
Dataset	`lakshmi25npathi_imdb-dataset-of-50k-movie-reviews_review_sentiment_train.jsonl`
Training examples	4102
Dataset artifact	`8c9516bb7750aaf508b195fc0c6df51599df0d9014508d0db75219b830157343`
Config artifact	`374b9cfd62da458a7dd3765373775d5ed1b569666658b2e0bb081ae2f44466c3`

The full Riva configuration is included in riva_config.yaml, and the run provenance is included in riva_provenance.json.

Evaluation

Phase	Metric	Value	Correct / Total	Step
baseline	classification_f1	0.0000	0 / 50	0
step	classification_f1	0.0303	1 / 50	10
step	classification_f1	0.8496	39 / 50	20
step	classification_f1	0.8730	40 / 50	30
step	classification_f1	0.8882	41 / 50	40
step	classification_f1	0.8882	41 / 50	50
final	classification_f1	0.8882	41 / 50	50
test	classification_f1	0.8738	405 / 500	50

Inference

python

from peft import AutoPeftModelForCausalLM
from transformers import AutoTokenizer

repo_id = "HassanJbr/gpt2-xl-classification-lora-riva"
tokenizer = AutoTokenizer.from_pretrained(repo_id)
model = AutoPeftModelForCausalLM.from_pretrained(repo_id, device_map="auto")

prompt = "Write your prompt here"
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=128)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Limitations

This model is an example checkpoint. Evaluate it on your own data and requirements before production use.
Behavior depends on the base model, the training dataset, and the task template used for this run.
Model and adapter use may be subject to the license and use restrictions of the base model and training data.

Riva

Train your own version: https://app.riva.run/import?hf=HassanJbr/gpt2-xl-classification-lora-riva

riva-labs/gpt2-xl-sentiment-analysis-riva

ancient-eagle