dad1909 commited on
Commit
db32a43
·
verified ·
1 Parent(s): e05a4e2

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +4 -3
app.py CHANGED
@@ -5,6 +5,7 @@ from trl import SFTTrainer
5
  from transformers import TrainingArguments
6
  from datasets import load_dataset
7
  import gradio as gr
 
8
  from huggingface_hub import HfApi
9
 
10
  max_seq_length = 4096
@@ -16,7 +17,7 @@ print("Starting model and tokenizer loading...")
16
 
17
  # Load the model and tokenizer
18
  model, tokenizer = FastLanguageModel.from_pretrained(
19
- model_name="unsloth/llama-3-8b-Instruct-bnb-4bit",
20
  max_seq_length=max_seq_length,
21
  dtype=dtype,
22
  load_in_4bit=load_in_4bit,
@@ -117,7 +118,7 @@ trainer = SFTTrainer(
117
  bf16=is_bfloat16_supported(),
118
  warmup_steps=5,
119
  logging_steps=10,
120
- max_steps=50,
121
  optim="adamw_8bit",
122
  weight_decay=0.01,
123
  lr_scheduler_type="linear",
@@ -137,7 +138,7 @@ print("Model saved successfully.")
137
 
138
  print("Pushing the model to the hub...")
139
  model.push_to_hub_merged(
140
- "CyberSentinel-initial",
141
  tokenizer,
142
  save_method="merged_16bit",
143
  token=hf_token
 
5
  from transformers import TrainingArguments
6
  from datasets import load_dataset
7
  import gradio as gr
8
+ import json
9
  from huggingface_hub import HfApi
10
 
11
  max_seq_length = 4096
 
17
 
18
  # Load the model and tokenizer
19
  model, tokenizer = FastLanguageModel.from_pretrained(
20
+ model_name="dad1909/CyberSentinel-initial",
21
  max_seq_length=max_seq_length,
22
  dtype=dtype,
23
  load_in_4bit=load_in_4bit,
 
118
  bf16=is_bfloat16_supported(),
119
  warmup_steps=5,
120
  logging_steps=10,
121
+ max_steps=200,
122
  optim="adamw_8bit",
123
  weight_decay=0.01,
124
  lr_scheduler_type="linear",
 
138
 
139
  print("Pushing the model to the hub...")
140
  model.push_to_hub_merged(
141
+ "CyberSentinel-1",
142
  tokenizer,
143
  save_method="merged_16bit",
144
  token=hf_token