Instructions to use tsavage68/chat_300_STEPS_03beta_5e7rate_CDPOSFT with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tsavage68/chat_300_STEPS_03beta_5e7rate_CDPOSFT with Transformers:

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

pipe = pipeline("text-generation", model="tsavage68/chat_300_STEPS_03beta_5e7rate_CDPOSFT")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tsavage68/chat_300_STEPS_03beta_5e7rate_CDPOSFT")
model = AutoModelForCausalLM.from_pretrained("tsavage68/chat_300_STEPS_03beta_5e7rate_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_300_STEPS_03beta_5e7rate_CDPOSFT with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tsavage68/chat_300_STEPS_03beta_5e7rate_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_300_STEPS_03beta_5e7rate_CDPOSFT",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tsavage68/chat_300_STEPS_03beta_5e7rate_CDPOSFT

SGLang

How to use tsavage68/chat_300_STEPS_03beta_5e7rate_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_300_STEPS_03beta_5e7rate_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_300_STEPS_03beta_5e7rate_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_300_STEPS_03beta_5e7rate_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_300_STEPS_03beta_5e7rate_CDPOSFT",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tsavage68/chat_300_STEPS_03beta_5e7rate_CDPOSFT with Docker Model Runner:
```
docker model run hf.co/tsavage68/chat_300_STEPS_03beta_5e7rate_CDPOSFT
```

chat_300_STEPS_03beta_5e7rate_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: 0.6793
Rewards/chosen: -0.0623
Rewards/rejected: -0.1009
Rewards/accuracies: 0.4945
Rewards/margins: 0.0386
Logps/rejected: -19.1383
Logps/chosen: -16.9623
Logits/rejected: -0.5997
Logits/chosen: -0.5996

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: 5e-07
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: 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
0.6938	0.0977	50	0.6926	0.0099	0.0081	0.4242	0.0018	-18.7749	-16.7215	-0.5989	-0.5987
0.6799	0.1953	100	0.6850	-0.0283	-0.0476	0.4769	0.0194	-18.9607	-16.8488	-0.6017	-0.6016
0.6902	0.2930	150	0.6833	-0.0843	-0.1125	0.4725	0.0281	-19.1769	-17.0358	-0.6005	-0.6004
0.6881	0.3906	200	0.6808	-0.0590	-0.0933	0.4835	0.0343	-19.1129	-16.9512	-0.5994	-0.5993
0.6919	0.4883	250	0.6812	-0.0623	-0.0971	0.4989	0.0348	-19.1258	-16.9623	-0.6003	-0.6001
0.6828	0.5859	300	0.6793	-0.0623	-0.1009	0.4945	0.0386	-19.1383	-16.9623	-0.5997	-0.5996

Framework versions

Transformers 4.40.1
Pytorch 2.0.0+cu117
Datasets 2.19.1
Tokenizers 0.19.1

Downloads last month: 6

Safetensors

Model size

7B params

Tensor type

F16

Model tree for tsavage68/chat_300_STEPS_03beta_5e7rate_CDPOSFT

Base model

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

Finetuned

tsavage68/chat_600STEPS_1e8rate_SFT

Finetuned

(23)

this model