Instructions to use tsavage68/Summary4500_L3_300steps_1e7rate_01beta_CSFTDPO with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tsavage68/Summary4500_L3_300steps_1e7rate_01beta_CSFTDPO with Transformers:

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

pipe = pipeline("text-generation", model="tsavage68/Summary4500_L3_300steps_1e7rate_01beta_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_300steps_1e7rate_01beta_CSFTDPO")
model = AutoModelForCausalLM.from_pretrained("tsavage68/Summary4500_L3_300steps_1e7rate_01beta_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_300steps_1e7rate_01beta_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_300steps_1e7rate_01beta_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_300steps_1e7rate_01beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tsavage68/Summary4500_L3_300steps_1e7rate_01beta_CSFTDPO

SGLang

How to use tsavage68/Summary4500_L3_300steps_1e7rate_01beta_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_300steps_1e7rate_01beta_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_300steps_1e7rate_01beta_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_300steps_1e7rate_01beta_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_300steps_1e7rate_01beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tsavage68/Summary4500_L3_300steps_1e7rate_01beta_CSFTDPO with Docker Model Runner:
```
docker model run hf.co/tsavage68/Summary4500_L3_300steps_1e7rate_01beta_CSFTDPO
```

Hyponatremia_L3_300steps_1e7rate_01beta_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: 0.1850
Rewards/chosen: 0.0851
Rewards/rejected: -1.5767
Rewards/accuracies: 0.9980
Rewards/margins: 1.6618
Logps/rejected: -148.9646
Logps/chosen: -83.3391
Logits/rejected: -1.1015
Logits/chosen: -1.0669

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-07
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: 300

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.6815	0.0112	50	0.6641	0.0083	-0.0520	0.8080	0.0603	-133.7178	-84.1071	-1.0987	-1.0686
0.4744	0.0224	100	0.4877	0.0371	-0.4326	0.9980	0.4697	-137.5237	-83.8192	-1.1002	-1.0687
0.1756	0.0336	150	0.2753	0.0731	-1.1076	0.9980	1.1807	-144.2736	-83.4591	-1.1015	-1.0683
0.1159	0.0448	200	0.2005	0.0847	-1.4796	0.9980	1.5642	-147.9930	-83.3433	-1.1016	-1.0674
0.1143	0.0559	250	0.1866	0.0833	-1.5665	0.9980	1.6498	-148.8622	-83.3566	-1.1017	-1.0670
0.0909	0.0671	300	0.1850	0.0851	-1.5767	0.9980	1.6618	-148.9646	-83.3391	-1.1015	-1.0669

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_300steps_1e7rate_01beta_CSFTDPO

Base model

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

Finetuned

tsavage68/Summary4500_L3_100steps_1e6rate_SFT

Finetuned

(14)

this model