Instructions to use PurCL/src_prober_codellama-13b-last1unfreeze with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use PurCL/src_prober_codellama-13b-last1unfreeze with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="PurCL/src_prober_codellama-13b-last1unfreeze")

# Load model directly
from transformers import AutoModelForSeq2SeqLM
model = AutoModelForSeq2SeqLM.from_pretrained("PurCL/src_prober_codellama-13b-last1unfreeze", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use PurCL/src_prober_codellama-13b-last1unfreeze with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "PurCL/src_prober_codellama-13b-last1unfreeze"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "PurCL/src_prober_codellama-13b-last1unfreeze",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/PurCL/src_prober_codellama-13b-last1unfreeze

SGLang

How to use PurCL/src_prober_codellama-13b-last1unfreeze with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "PurCL/src_prober_codellama-13b-last1unfreeze" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "PurCL/src_prober_codellama-13b-last1unfreeze",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "PurCL/src_prober_codellama-13b-last1unfreeze" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "PurCL/src_prober_codellama-13b-last1unfreeze",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use PurCL/src_prober_codellama-13b-last1unfreeze with Docker Model Runner:
```
docker model run hf.co/PurCL/src_prober_codellama-13b-last1unfreeze
```

src_prober_codellama-13b-last1unfreeze

This model is a fine-tuned version of on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.6267

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 5e-05
train_batch_size: 4
eval_batch_size: 8
seed: 42
distributed_type: multi-GPU
num_devices: 4
gradient_accumulation_steps: 4
total_train_batch_size: 64
total_eval_batch_size: 32
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 1000
num_epochs: 5
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.7443	0.12	500	0.7429
0.6851	0.24	1000	0.7170
0.6723	0.36	1500	0.6912
0.6605	0.48	2000	0.6730
0.6475	0.6	2500	0.6643
0.6419	0.72	3000	0.6584
0.6307	0.85	3500	0.6532
0.6167	0.97	4000	0.6495
0.6272	1.09	4500	0.6477
0.6002	1.21	5000	0.6445
0.6303	1.33	5500	0.6429
0.6405	1.45	6000	0.6421
0.6041	1.57	6500	0.6387
0.5912	1.69	7000	0.6370
0.6121	1.81	7500	0.6360
0.613	1.93	8000	0.6344
0.6126	2.05	8500	0.6338
0.5932	2.17	9000	0.6344
0.5927	2.3	9500	0.6332
0.5883	2.42	10000	0.6317
0.6023	2.54	10500	0.6308
0.5898	2.66	11000	0.6311
0.576	2.78	11500	0.6291
0.5699	2.9	12000	0.6291
0.6093	3.02	12500	0.6290
0.5754	3.14	13000	0.6292
0.6294	3.26	13500	0.6282
0.591	3.38	14000	0.6283
0.599	3.5	14500	0.6273
0.5933	3.62	15000	0.6281
0.565	3.75	15500	0.6268
0.5884	3.87	16000	0.6267
0.5809	3.99	16500	0.6266
0.5618	4.11	17000	0.6271
0.5749	4.23	17500	0.6274
0.577	4.35	18000	0.6268
0.5947	4.47	18500	0.6267
0.5902	4.59	19000	0.6268
0.5869	4.71	19500	0.6268
0.5829	4.83	20000	0.6268
0.5587	4.95	20500	0.6267

Framework versions

Transformers 4.39.3
Pytorch 2.2.0
Datasets 2.17.0
Tokenizers 0.15.2

Downloads last month: 8

Safetensors

Model size

0.2B params

Tensor type

F32

Collection including PurCL/src_prober_codellama-13b-last1unfreeze

ProRec

Collection

Data and Models for "Source Code Foundation Models are Transferable Binary Analysis Knowledge Bases" • 15 items • Updated Jun 3, 2024