Instructions to use tsavage68/chat_700_STEPS_03beta_1e6rate_CDPOSFT with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tsavage68/chat_700_STEPS_03beta_1e6rate_CDPOSFT with Transformers:

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

pipe = pipeline("text-generation", model="tsavage68/chat_700_STEPS_03beta_1e6rate_CDPOSFT")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tsavage68/chat_700_STEPS_03beta_1e6rate_CDPOSFT")
model = AutoModelForCausalLM.from_pretrained("tsavage68/chat_700_STEPS_03beta_1e6rate_CDPOSFT", 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_700_STEPS_03beta_1e6rate_CDPOSFT with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tsavage68/chat_700_STEPS_03beta_1e6rate_CDPOSFT"
# 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_700_STEPS_03beta_1e6rate_CDPOSFT",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tsavage68/chat_700_STEPS_03beta_1e6rate_CDPOSFT

SGLang

How to use tsavage68/chat_700_STEPS_03beta_1e6rate_CDPOSFT 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_700_STEPS_03beta_1e6rate_CDPOSFT" \
    --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_700_STEPS_03beta_1e6rate_CDPOSFT",
		"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_700_STEPS_03beta_1e6rate_CDPOSFT" \
        --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_700_STEPS_03beta_1e6rate_CDPOSFT",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tsavage68/chat_700_STEPS_03beta_1e6rate_CDPOSFT with Docker Model Runner:
```
docker model run hf.co/tsavage68/chat_700_STEPS_03beta_1e6rate_CDPOSFT
```

chat_700_STEPS_03beta_1e6rate_CDPOSFT

This model is a fine-tuned version of tsavage68/chat_600STEPS_1e8rate_SFT on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.6706
Rewards/chosen: -0.2188
Rewards/rejected: -0.3671
Rewards/accuracies: 0.5143
Rewards/margins: 0.1484
Logps/rejected: -20.0258
Logps/chosen: -17.4839
Logits/rejected: -0.6007
Logits/chosen: -0.6005

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: 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
0.6903	0.0977	50	0.6898	0.0339	0.0260	0.4264	0.0078	-18.7152	-16.6418	-0.6000	-0.5999
0.6568	0.1953	100	0.6714	-0.1082	-0.1762	0.5099	0.0680	-19.3893	-17.1151	-0.6152	-0.6151
0.7125	0.2930	150	0.6838	-0.1101	-0.1755	0.4791	0.0653	-19.3869	-17.1217	-0.5952	-0.5950
0.7095	0.3906	200	0.6820	-0.1564	-0.2410	0.5055	0.0846	-19.6053	-17.2759	-0.5844	-0.5842
0.7264	0.4883	250	0.6859	-0.0974	-0.1989	0.4967	0.1016	-19.4651	-17.0792	-0.5778	-0.5776
0.6767	0.5859	300	0.6737	-0.2009	-0.3435	0.5121	0.1426	-19.9470	-17.4243	-0.6046	-0.6044
0.6546	0.6836	350	0.6776	-0.2753	-0.4068	0.5033	0.1316	-20.1581	-17.6722	-0.5869	-0.5867
0.6473	0.7812	400	0.6697	-0.2700	-0.4199	0.5209	0.1499	-20.2016	-17.6546	-0.6084	-0.6082
0.68	0.8789	450	0.6720	-0.2073	-0.3505	0.5121	0.1432	-19.9703	-17.4455	-0.5885	-0.5883
0.6626	0.9766	500	0.6726	-0.2140	-0.3584	0.5099	0.1444	-19.9967	-17.4681	-0.5948	-0.5946
0.3861	1.0742	550	0.6702	-0.2078	-0.3569	0.5209	0.1492	-19.9917	-17.4471	-0.5992	-0.5990
0.4031	1.1719	600	0.6720	-0.2186	-0.3641	0.5121	0.1455	-20.0158	-17.4834	-0.6004	-0.6002
0.4139	1.2695	650	0.6703	-0.2170	-0.3648	0.5121	0.1478	-20.0179	-17.4778	-0.6006	-0.6004
0.3251	1.3672	700	0.6706	-0.2188	-0.3671	0.5143	0.1484	-20.0258	-17.4839	-0.6007	-0.6005

Framework versions

Transformers 4.40.1
Pytorch 2.0.0+cu117
Datasets 2.19.0
Tokenizers 0.19.1

Downloads last month: 6

Safetensors

Model size

7B params

Tensor type

F16

Model tree for tsavage68/chat_700_STEPS_03beta_1e6rate_CDPOSFT

Base model

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

Finetuned

tsavage68/chat_600STEPS_1e8rate_SFT

Finetuned

(23)

this model