Instructions to use tsavage68/Transaminitis_L3_350steps_1e7rate_05beta_CSFTDPO with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tsavage68/Transaminitis_L3_350steps_1e7rate_05beta_CSFTDPO with Transformers:

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

pipe = pipeline("text-generation", model="tsavage68/Transaminitis_L3_350steps_1e7rate_05beta_CSFTDPO")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tsavage68/Transaminitis_L3_350steps_1e7rate_05beta_CSFTDPO")
model = AutoModelForCausalLM.from_pretrained("tsavage68/Transaminitis_L3_350steps_1e7rate_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/Transaminitis_L3_350steps_1e7rate_05beta_CSFTDPO with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tsavage68/Transaminitis_L3_350steps_1e7rate_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/Transaminitis_L3_350steps_1e7rate_05beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tsavage68/Transaminitis_L3_350steps_1e7rate_05beta_CSFTDPO

SGLang

How to use tsavage68/Transaminitis_L3_350steps_1e7rate_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/Transaminitis_L3_350steps_1e7rate_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/Transaminitis_L3_350steps_1e7rate_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/Transaminitis_L3_350steps_1e7rate_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/Transaminitis_L3_350steps_1e7rate_05beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tsavage68/Transaminitis_L3_350steps_1e7rate_05beta_CSFTDPO with Docker Model Runner:
```
docker model run hf.co/tsavage68/Transaminitis_L3_350steps_1e7rate_05beta_CSFTDPO
```

Transaminitis_L3_350steps_1e7rate_05beta_CSFTDPO

This model is a fine-tuned version of tsavage68/Transaminitis_L3_1000rate_1e7_SFT on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.5677
Rewards/chosen: 0.0947
Rewards/rejected: -0.2002
Rewards/accuracies: 0.8600
Rewards/margins: 0.2949
Logps/rejected: -18.9551
Logps/chosen: -18.3449
Logits/rejected: -1.0739
Logits/chosen: -1.0723

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: 2
eval_batch_size: 1
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 4
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 100
training_steps: 350

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.6857	0.2	25	0.6895	-0.0206	-0.0288	0.5100	0.0082	-18.6123	-18.5753	-1.0653	-1.0641
0.6912	0.4	50	0.6888	-0.1407	-0.1512	0.5300	0.0106	-18.8572	-18.8156	-1.0675	-1.0663
0.6956	0.6	75	0.6978	-0.1002	-0.1011	0.4600	0.0008	-18.7568	-18.7347	-1.0682	-1.0669
0.6647	0.8	100	0.7297	-0.2211	-0.2283	0.4600	0.0071	-19.0112	-18.9765	-1.0701	-1.0690
0.7239	1.0	125	0.6908	-0.6506	-0.7800	0.5400	0.1293	-20.1146	-19.8355	-1.0728	-1.0716
0.6533	1.2	150	0.6792	0.0691	-0.0036	0.4700	0.0728	-18.5620	-18.3960	-1.0696	-1.0682
0.6223	1.4	175	0.6196	-0.1328	-0.2981	0.7800	0.1652	-19.1508	-18.7999	-1.0734	-1.0721
0.6026	1.6	200	0.5921	-0.1823	-0.4363	0.7300	0.2539	-19.4273	-18.8989	-1.0736	-1.0723
0.5946	1.8	225	0.5779	0.0165	-0.2513	0.8300	0.2678	-19.0573	-18.5012	-1.0748	-1.0732
0.5438	2.0	250	0.5756	0.0271	-0.2507	0.8200	0.2778	-19.0561	-18.4800	-1.0745	-1.0731
0.5717	2.2	275	0.5683	0.0778	-0.2143	0.8500	0.2921	-18.9833	-18.3785	-1.0744	-1.0730
0.5337	2.4	300	0.5698	0.0926	-0.1967	0.8600	0.2894	-18.9482	-18.3489	-1.0749	-1.0735
0.5534	2.6	325	0.5667	0.1026	-0.1939	0.8600	0.2965	-18.9425	-18.3291	-1.0738	-1.0723
0.5358	2.8	350	0.5677	0.0947	-0.2002	0.8600	0.2949	-18.9551	-18.3449	-1.0739	-1.0723

Framework versions

Transformers 4.40.2
Pytorch 2.0.0+cu117
Datasets 2.19.1
Tokenizers 0.19.1

Downloads last month: 5

Safetensors

Model size

8B params

Tensor type

F16

Model tree for tsavage68/Transaminitis_L3_350steps_1e7rate_05beta_CSFTDPO

Base model

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

Finetuned

tsavage68/Transaminitis_L3_1000rate_1e7_SFT

Finetuned

(17)

this model