Instructions to use tsavage68/Summary4500_L3_1000steps_1e6rate_05beta_CSFTDPO with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tsavage68/Summary4500_L3_1000steps_1e6rate_05beta_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_05beta_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_05beta_CSFTDPO")
model = AutoModelForCausalLM.from_pretrained("tsavage68/Summary4500_L3_1000steps_1e6rate_05beta_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_05beta_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_05beta_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_05beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tsavage68/Summary4500_L3_1000steps_1e6rate_05beta_CSFTDPO

SGLang

How to use tsavage68/Summary4500_L3_1000steps_1e6rate_05beta_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_05beta_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_05beta_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_05beta_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_05beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tsavage68/Summary4500_L3_1000steps_1e6rate_05beta_CSFTDPO with Docker Model Runner:
```
docker model run hf.co/tsavage68/Summary4500_L3_1000steps_1e6rate_05beta_CSFTDPO
```

Hyponatremia_L3_1000steps_1e6rate_05beta_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: 1.0176
Rewards/rejected: -20.0926
Rewards/accuracies: 0.9980
Rewards/margins: 21.1103
Logps/rejected: -173.3826
Logps/chosen: -82.1545
Logits/rejected: -1.0918
Logits/chosen: -1.0524

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
0.0003	0.0112	50	0.0023	0.5359	-8.1120	0.9980	8.6479	-149.4213	-83.1180	-1.1012	-1.0676
0.0	0.0224	100	0.0016	0.1709	-10.9251	0.9980	11.0960	-155.0475	-83.8480	-1.1026	-1.0673
0.0	0.0336	150	0.0014	-0.1278	-13.7945	0.9980	13.6667	-160.7863	-84.4453	-1.1022	-1.0664
0.0	0.0448	200	0.0014	-0.0574	-14.6683	0.9980	14.6109	-162.5339	-84.3046	-1.1016	-1.0657
0.0	0.0559	250	0.0014	0.3311	-15.4389	0.9980	15.7700	-164.0751	-83.5275	-1.0992	-1.0628
0.0	0.0671	300	0.0014	0.3433	-15.4472	0.9980	15.7905	-164.0917	-83.5031	-1.0990	-1.0626
0.0	0.0783	350	0.0014	0.4029	-17.0508	0.9980	17.4537	-167.2989	-83.3839	-1.1027	-1.0639
0.0	0.0895	400	0.0014	0.3792	-17.1575	0.9980	17.5367	-167.5124	-83.4315	-1.1026	-1.0637
0.0	0.1007	450	0.0014	0.4159	-17.1507	0.9980	17.5667	-167.4988	-83.3579	-1.1033	-1.0647
0.0	0.1119	500	0.0014	0.6555	-18.5577	0.9980	19.2132	-170.3127	-82.8788	-1.0977	-1.0583
0.0	0.1231	550	0.0014	0.9891	-20.0773	0.9980	21.0664	-173.3519	-82.2115	-1.0934	-1.0539
0.0	0.1343	600	0.0014	0.9858	-20.0819	0.9980	21.0676	-173.3611	-82.2182	-1.0935	-1.0539
0.0	0.1454	650	0.0014	0.9858	-20.0819	0.9980	21.0676	-173.3611	-82.2182	-1.0935	-1.0539
0.0	0.1566	700	0.0014	0.9752	-20.1001	0.9980	21.0753	-173.3975	-82.2393	-1.0933	-1.0536
0.0	0.1678	750	0.0014	0.9974	-20.1078	0.9980	21.1052	-173.4129	-82.1949	-1.0923	-1.0527
0.0	0.1790	800	0.0014	1.0079	-20.1039	0.9980	21.1118	-173.4052	-82.1740	-1.0923	-1.0528
0.0	0.1902	850	0.0014	1.0134	-20.1134	0.9980	21.1268	-173.4241	-82.1630	-1.0920	-1.0524
0.0	0.2014	900	0.0014	1.0201	-20.0711	0.9980	21.0912	-173.3395	-82.1496	-1.0918	-1.0524
0.0	0.2126	950	0.0014	1.0208	-20.0898	0.9980	21.1107	-173.3770	-82.1481	-1.0918	-1.0524
0.0	0.2238	1000	0.0014	1.0176	-20.0926	0.9980	21.1103	-173.3826	-82.1545	-1.0918	-1.0524

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_05beta_CSFTDPO

Base model

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

Finetuned

tsavage68/Summary4500_L3_100steps_1e6rate_SFT

Finetuned

(14)

this model