Aditya369 — Qwen3-4B emotion classifier (merged)

by AGK FIRE INC

A Qwen3-4B model fine-tuned for emotion classification on the dair-ai/emotion dataset (20k English tweets/short texts). This is the merged version — LoRA weights fused into the base model, still 4-bit quantized. For the standalone LoRA adapter, see agk4444/aditya369-lora.

Labels

id	emotion
0	sadness
1	joy
2	love
3	anger
4	fear
5	surprise

Usage

from transformers import AutoModelForSequenceClassification, AutoTokenizer

model = AutoModelForSequenceClassification.from_pretrained("agk4444/aditya369")
tok = AutoTokenizer.from_pretrained("agk4444/aditya369")

labels = ["sadness", "joy", "love", "anger", "fear", "surprise"]

text = "I am so happy today!"
inputs = tok(text, return_tensors="pt")
pred = model(**inputs).logits.argmax(-1).item()
print(labels[pred])  # joy

Requires bitsandbytes (4-bit weights) and transformers.

Evaluation

Test-set results on dair-ai/emotion (2k held-out examples):

metric	score
accuracy	0.9350
macro F1	0.8907
weighted F1	0.9341
test loss	0.2754

Per-class results (test set)

emotion	precision	recall	f1	n
sadness	0.968	0.979	0.973	581
joy	0.941	0.964	0.952	695
love	0.856	0.786	0.820	159
anger	0.942	0.942	0.942	275
fear	0.905	0.893	0.899	224
surprise	0.810	0.712	0.758	66

Training

• Base: Qwen/Qwen3-4B, loaded in 4-bit (bitsandbytes)
• Method: LoRA (r=16), fused into base weights after training
• Data: dair-ai/emotion — 16k train / 2k val / 2k test
• 3 epochs (~1,500 steps), gradient checkpointing on
• Trained on a single T4 (Kaggle)

Limitations

• English short-form text (tweets); performance on long documents or other languages is untested.
• Six coarse emotions only — no intensity scores, no mixed-emotion output.
• Inherits the biases of the base model and the tweet dataset.

---

© 2026 AGK FIRE INC. Released under Apache 2.0.