Instructions to use FleetAI/fleet-sft-overfit-github-Qwen3-32B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use FleetAI/fleet-sft-overfit-github-Qwen3-32B with Transformers:

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

pipe = pipeline("text-generation", model="FleetAI/fleet-sft-overfit-github-Qwen3-32B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("FleetAI/fleet-sft-overfit-github-Qwen3-32B")
model = AutoModelForCausalLM.from_pretrained("FleetAI/fleet-sft-overfit-github-Qwen3-32B", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use FleetAI/fleet-sft-overfit-github-Qwen3-32B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "FleetAI/fleet-sft-overfit-github-Qwen3-32B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "FleetAI/fleet-sft-overfit-github-Qwen3-32B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/FleetAI/fleet-sft-overfit-github-Qwen3-32B

SGLang

How to use FleetAI/fleet-sft-overfit-github-Qwen3-32B 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 "FleetAI/fleet-sft-overfit-github-Qwen3-32B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "FleetAI/fleet-sft-overfit-github-Qwen3-32B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "FleetAI/fleet-sft-overfit-github-Qwen3-32B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "FleetAI/fleet-sft-overfit-github-Qwen3-32B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use FleetAI/fleet-sft-overfit-github-Qwen3-32B with Docker Model Runner:
```
docker model run hf.co/FleetAI/fleet-sft-overfit-github-Qwen3-32B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

fleet-sft-full

This model is a fine-tuned version of Qwen/Qwen3-32B on the fleet_trajectories_train dataset. It achieves the following results on the evaluation set:

Loss: 1.8278

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: 2e-05
train_batch_size: 1
eval_batch_size: 1
seed: 42
distributed_type: multi-GPU
num_devices: 8
total_train_batch_size: 8
total_eval_batch_size: 8
optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_ratio: 0.05
num_epochs: 50.0

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0	0	1.4682
1.3518	0.7143	5	1.3431
0.9051	1.4286	10	1.2059
0.8625	2.1429	15	1.1277
0.6971	2.8571	20	1.1145
0.4243	3.5714	25	1.1612
0.1901	4.2857	30	1.3399
0.1171	5.0	35	1.3959
0.0593	5.7143	40	1.4992
0.0265	6.4286	45	1.5967
0.0229	7.1429	50	1.6269
0.0167	7.8571	55	1.6505
0.0119	8.5714	60	1.6626
0.0094	9.2857	65	1.6874
0.0089	10.0	70	1.7035
0.0062	10.7143	75	1.7053
0.0072	11.4286	80	1.7082
0.008	12.1429	85	1.6972
0.006	12.8571	90	1.6969
0.0044	13.5714	95	1.7049
0.0036	14.2857	100	1.7229
0.0035	15.0	105	1.7421
0.0026	15.7143	110	1.7550
0.0023	16.4286	115	1.7617
0.0021	17.1429	120	1.7656
0.0025	17.8571	125	1.7683
0.0018	18.5714	130	1.7750
0.002	19.2857	135	1.7667
0.0036	20.0	140	1.7492
0.0025	20.7143	145	1.7378
0.0017	21.4286	150	1.7389
0.0016	22.1429	155	1.7510
0.0016	22.8571	160	1.7623
0.0014	23.5714	165	1.7705
0.0013	24.2857	170	1.7751
0.0015	25.0	175	1.7802
0.0011	25.7143	180	1.7830
0.0012	26.4286	185	1.7873
0.0011	27.1429	190	1.7919
0.0012	27.8571	195	1.7959
0.0012	28.5714	200	1.7993
0.001	29.2857	205	1.8018
0.0012	30.0	210	1.8040
0.001	30.7143	215	1.8073
0.001	31.4286	220	1.8092
0.0014	32.1429	225	1.8116
0.0011	32.8571	230	1.8135
0.001	33.5714	235	1.8141
0.0011	34.2857	240	1.8167
0.0009	35.0	245	1.8182
0.001	35.7143	250	1.8190
0.0011	36.4286	255	1.8204
0.0012	37.1429	260	1.8216
0.0009	37.8571	265	1.8221
0.001	38.5714	270	1.8223
0.0013	39.2857	275	1.8238
0.0011	40.0	280	1.8247
0.0009	40.7143	285	1.8251
0.0011	41.4286	290	1.8253
0.001	42.1429	295	1.8262
0.001	42.8571	300	1.8267
0.0011	43.5714	305	1.8267
0.0012	44.2857	310	1.8272
0.0009	45.0	315	1.8278
0.0008	45.7143	320	1.8276
0.0009	46.4286	325	1.8282
0.001	47.1429	330	1.8282
0.001	47.8571	335	1.8279
0.0008	48.5714	340	1.8282
0.0012	49.2857	345	1.8281
0.001	50.0	350	1.8278

Framework versions

Transformers 4.52.4
Pytorch 2.10.0+cu128
Datasets 3.6.0
Tokenizers 0.21.1

Downloads last month: 7

Safetensors

Model size

33B params

Tensor type

BF16

Model tree for FleetAI/fleet-sft-overfit-github-Qwen3-32B

Base model

Qwen/Qwen3-32B

Finetuned

(522)

this model

Quantizations

1 model