Upload 7 files

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -1,55 +1,44 @@
----
-language:
-- en
-license: apache-2.0
-tags:
-- peft
-- lora
-- adapter
-- emotion-classification
-base_model: unsloth/Meta-Llama-3.1-8B-Instruct
-library_name: peft
----
+# Fine-Tuned Emotion Classification Model
 
-# Mirari - LoRA Emotion Classification Adapter
+## Model Information
+- **Base Model**: unsloth/Meta-Llama-3.1-8B-Instruct
+- **Training Method**: LoRA (Low-Rank Adaptation)
+- **LoRA Rank**: 32
+- **Training Samples**: 56,400
+- **Datasets Used**: GoEmotions, Emotion, TweetEval
 
-This is a **LoRA (Low-Rank Adaptation)** adapter model for emotion classification.
-
-## Model Type
-- **Type:** LoRA Adapter
-- **Base Model:** Meta-Llama-3.1-8B-Instruct (or similar)
-- **Task:** Emotion Classification
-- **Framework:** PEFT (Parameter-Efficient Fine-Tuning)
-
-## Usage
-
-This is an adapter model and requires a base model to function.
+## How to Load This Model
 
 ```python
-from transformers import AutoModelForCausalLM, AutoTokenizer
-from peft import PeftModel
+from unsloth import FastLanguageModel
 
-# Load base model (adjust to your actual base model)
-base_model_name = "unsloth/Meta-Llama-3.1-8B-Instruct"
-base_model = AutoModelForCausalLM.from_pretrained(base_model_name)
-tokenizer = AutoTokenizer.from_pretrained(base_model_name)
+# Load the fine-tuned model
+model, tokenizer = FastLanguageModel.from_pretrained(
+    model_name="emotion_model_finetuned",
+    max_seq_length=2048,
+    dtype=None,
+    load_in_4bit=True,
+)
 
-# Load LoRA adapter
-model = PeftModel.from_pretrained(base_model, "RockyBai/Mirari")
+# Enable inference mode
+FastLanguageModel.for_inference(model)
 
 # Use the model
-text = "I am so happy today!"
-inputs = tokenizer(text, return_tensors="pt")
-outputs = model.generate(**inputs, max_new_tokens=50)
-print(tokenizer.decode(outputs[0]))
+prompt = """<|im_start|>system
+You are a compassionate mental health support assistant.<|im_end|>
+<|im_start|>user
+I'm feeling anxious about tomorrow.<|im_end|>
+<|im_start|>assistant
+"""
+
+inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
+outputs = model.generate(**inputs, max_new_tokens=128)
+response = tokenizer.decode(outputs[0], skip_special_tokens=True)
+print(response)
 ```
 
-## Training Details
-
-- Adapter size: ~320MB
-- Training method: LoRA (Low-Rank Adaptation)
-- Task: Emotion classification
-
-## Notes
-
-This model requires the base model to be loaded first, then this adapter is applied on top of it.
+## Files Included
+- `adapter_config.json` - LoRA adapter configuration
+- `adapter_model.safetensors` - Fine-tuned weights
+- `tokenizer.json` - Tokenizer files
+- `training_config.json` - Training hyperparameters

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b9e4e7fb171f92fd137b777cc2714bf87d11576700a1dcd7a399e7bbe39537b
+size 17209920