Instructions to use tsavage68/chat_700STEPS_1e4rate_01beta_DPO with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tsavage68/chat_700STEPS_1e4rate_01beta_DPO with Transformers:

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

pipe = pipeline("text-generation", model="tsavage68/chat_700STEPS_1e4rate_01beta_DPO")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tsavage68/chat_700STEPS_1e4rate_01beta_DPO")
model = AutoModelForCausalLM.from_pretrained("tsavage68/chat_700STEPS_1e4rate_01beta_DPO", 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/chat_700STEPS_1e4rate_01beta_DPO with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tsavage68/chat_700STEPS_1e4rate_01beta_DPO"
# 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/chat_700STEPS_1e4rate_01beta_DPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tsavage68/chat_700STEPS_1e4rate_01beta_DPO

SGLang

How to use tsavage68/chat_700STEPS_1e4rate_01beta_DPO 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/chat_700STEPS_1e4rate_01beta_DPO" \
    --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/chat_700STEPS_1e4rate_01beta_DPO",
		"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/chat_700STEPS_1e4rate_01beta_DPO" \
        --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/chat_700STEPS_1e4rate_01beta_DPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tsavage68/chat_700STEPS_1e4rate_01beta_DPO with Docker Model Runner:
```
docker model run hf.co/tsavage68/chat_700STEPS_1e4rate_01beta_DPO
```

chat_700STEPS_1e4rate

This model is a fine-tuned version of meta-llama/Llama-2-7b-chat-hf on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 1.1848
Rewards/chosen: -4.4236
Rewards/rejected: -4.3538
Rewards/accuracies: 0.4000
Rewards/margins: -0.0698
Logps/rejected: -62.3289
Logps/chosen: -60.9807
Logits/rejected: -4.5000
Logits/chosen: -4.5000

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: 0.0001
train_batch_size: 4
eval_batch_size: 1
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 8
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 100
training_steps: 700

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
1.2329	0.1	50	1.6243	-8.0715	-8.0351	0.4176	-0.0364	-99.1426	-97.4598	-0.6161	-0.6164
1.3399	0.2	100	1.2323	-4.6270	-4.6104	0.3978	-0.0167	-64.8951	-63.0151	-2.3014	-2.3014
1.316	0.29	150	1.2017	-4.3807	-4.3323	0.4000	-0.0484	-62.1140	-60.5517	-2.9891	-2.9891
1.2778	0.39	200	1.1891	-4.3216	-4.2623	0.4044	-0.0593	-61.4138	-59.9605	-3.4116	-3.4116
1.0721	0.49	250	1.1847	-4.3234	-4.2565	0.4110	-0.0669	-61.3561	-59.9788	-3.7977	-3.7977
1.3775	0.59	300	1.1896	-4.3481	-4.2745	0.4176	-0.0737	-61.5360	-60.2260	-4.0911	-4.0911
1.3232	0.68	350	1.1818	-4.3486	-4.2846	0.4088	-0.0640	-61.6375	-60.2310	-4.2490	-4.2490
1.2476	0.78	400	1.1789	-4.3705	-4.3105	0.4044	-0.0600	-61.8961	-60.4495	-4.1920	-4.1920
1.3082	0.88	450	1.1766	-4.3831	-4.3296	0.4000	-0.0535	-62.0870	-60.5753	-4.2348	-4.2348
1.1007	0.98	500	1.1762	-4.4022	-4.3470	0.4000	-0.0552	-62.2610	-60.7662	-4.2010	-4.2010
0.8786	1.07	550	1.1811	-4.4150	-4.3527	0.4000	-0.0623	-62.3185	-60.8948	-4.3351	-4.3351
1.3113	1.17	600	1.1842	-4.4174	-4.3487	0.4000	-0.0687	-62.2785	-60.9186	-4.4858	-4.4858
0.9783	1.27	650	1.1850	-4.4234	-4.3533	0.4000	-0.0701	-62.3242	-60.9783	-4.4997	-4.4997
1.3696	1.37	700	1.1848	-4.4236	-4.3538	0.4000	-0.0698	-62.3289	-60.9807	-4.5000	-4.5000

Framework versions

Transformers 4.37.2
Pytorch 2.0.0+cu117
Datasets 2.17.0
Tokenizers 0.15.2

Downloads last month: 5

Safetensors

Model size

7B params

Tensor type

F16

Model tree for tsavage68/chat_700STEPS_1e4rate_01beta_DPO

Base model

meta-llama/Llama-2-7b-chat-hf

Finetuned

(552)

this model