Instructions to use tsavage68/Summary4500_L3_1000steps_1e6rate_03beta_CSFTDPO with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tsavage68/Summary4500_L3_1000steps_1e6rate_03beta_CSFTDPO with Transformers:

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

pipe = pipeline("text-generation", model="tsavage68/Summary4500_L3_1000steps_1e6rate_03beta_CSFTDPO")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tsavage68/Summary4500_L3_1000steps_1e6rate_03beta_CSFTDPO")
model = AutoModelForCausalLM.from_pretrained("tsavage68/Summary4500_L3_1000steps_1e6rate_03beta_CSFTDPO", 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 tsavage68/Summary4500_L3_1000steps_1e6rate_03beta_CSFTDPO with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tsavage68/Summary4500_L3_1000steps_1e6rate_03beta_CSFTDPO"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "tsavage68/Summary4500_L3_1000steps_1e6rate_03beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tsavage68/Summary4500_L3_1000steps_1e6rate_03beta_CSFTDPO

SGLang

How to use tsavage68/Summary4500_L3_1000steps_1e6rate_03beta_CSFTDPO 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 "tsavage68/Summary4500_L3_1000steps_1e6rate_03beta_CSFTDPO" \
    --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": "tsavage68/Summary4500_L3_1000steps_1e6rate_03beta_CSFTDPO",
		"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 "tsavage68/Summary4500_L3_1000steps_1e6rate_03beta_CSFTDPO" \
        --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": "tsavage68/Summary4500_L3_1000steps_1e6rate_03beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tsavage68/Summary4500_L3_1000steps_1e6rate_03beta_CSFTDPO with Docker Model Runner:
```
docker model run hf.co/tsavage68/Summary4500_L3_1000steps_1e6rate_03beta_CSFTDPO
```

Hyponatremia_L3_1000steps_1e6rate_03beta_CSFTDPO

This model is a fine-tuned version of tsavage68/Summary4500_L3_100steps_1e6rate_SFT on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.0014
Rewards/chosen: -0.1633
Rewards/rejected: -19.7875
Rewards/accuracies: 0.9980
Rewards/margins: 19.6243
Logps/rejected: -199.1558
Logps/chosen: -84.7341
Logits/rejected: -1.0916
Logits/chosen: -1.0431

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: 1e-06
train_batch_size: 1
eval_batch_size: 1
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 100
training_steps: 1000

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.0011	0.0112	50	0.0032	0.4589	-7.1323	0.9980	7.5912	-156.9717	-82.6602	-1.1014	-1.0652
0.0	0.0224	100	0.0017	0.0259	-10.1984	0.9980	10.2243	-167.1920	-84.1034	-1.1010	-1.0621
0.0	0.0336	150	0.0015	-0.2730	-12.2233	0.9980	11.9503	-173.9416	-85.0998	-1.1007	-1.0606
0.0	0.0448	200	0.0014	-0.2383	-14.0974	0.9980	13.8592	-180.1888	-84.9840	-1.0957	-1.0547
0.0	0.0559	250	0.0014	-0.4961	-16.6298	0.9980	16.1337	-188.6300	-85.8433	-1.0906	-1.0485
0.0	0.0671	300	0.0014	-0.4855	-16.6491	0.9980	16.1636	-188.6945	-85.8082	-1.0906	-1.0484
0.0	0.0783	350	0.0014	-0.4651	-18.0207	0.9980	17.5556	-193.2663	-85.7401	-1.0930	-1.0475
0.0	0.0895	400	0.0014	-0.4705	-18.0770	0.9980	17.6065	-193.4542	-85.7582	-1.0925	-1.0469
0.0	0.1007	450	0.0014	-0.4749	-18.1128	0.9980	17.6379	-193.5734	-85.7727	-1.0927	-1.0470
0.0	0.1119	500	0.0014	-0.4497	-18.3137	0.9980	17.8641	-194.2431	-85.6886	-1.0920	-1.0462
0.0	0.1231	550	0.0014	-0.1952	-19.8131	0.9980	19.6179	-199.2410	-84.8404	-1.0929	-1.0442
0.0	0.1343	600	0.0014	-0.1956	-19.8283	0.9980	19.6327	-199.2916	-84.8418	-1.0929	-1.0442
0.0	0.1454	650	0.0014	-0.1887	-19.8240	0.9980	19.6353	-199.2772	-84.8187	-1.0930	-1.0444
0.0	0.1566	700	0.0014	-0.1862	-19.8230	0.9980	19.6368	-199.2740	-84.8106	-1.0930	-1.0443
0.0	0.1678	750	0.0014	-0.1676	-19.7855	0.9980	19.6180	-199.1491	-84.7483	-1.0918	-1.0432
0.0	0.1790	800	0.0014	-0.1614	-19.7862	0.9980	19.6248	-199.1514	-84.7279	-1.0917	-1.0430
0.0	0.1902	850	0.0014	-0.1737	-19.8108	0.9980	19.6371	-199.2332	-84.7688	-1.0916	-1.0433
0.0	0.2014	900	0.0014	-0.1638	-19.8003	0.9980	19.6364	-199.1983	-84.7359	-1.0916	-1.0432
0.0	0.2126	950	0.0014	-0.1645	-19.7862	0.9980	19.6217	-199.1513	-84.7380	-1.0916	-1.0431
0.0	0.2238	1000	0.0014	-0.1633	-19.7875	0.9980	19.6243	-199.1558	-84.7341	-1.0916	-1.0431

Framework versions

Transformers 4.42.4
Pytorch 2.0.0+cu117
Datasets 2.20.0
Tokenizers 0.19.1

Downloads last month: 6

Safetensors

Model size

8B params

Tensor type

F16

Model tree for tsavage68/Summary4500_L3_1000steps_1e6rate_03beta_CSFTDPO

Base model

meta-llama/Meta-Llama-3-8B-Instruct

Finetuned

tsavage68/Summary4500_L3_100steps_1e6rate_SFT

Finetuned

(14)

this model