Instructions to use DatPySci/pythia-410m-sft-full with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use DatPySci/pythia-410m-sft-full with Transformers:

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

pipe = pipeline("text-generation", model="DatPySci/pythia-410m-sft-full")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("DatPySci/pythia-410m-sft-full")
model = AutoModelForCausalLM.from_pretrained("DatPySci/pythia-410m-sft-full")
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

vLLM

How to use DatPySci/pythia-410m-sft-full with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "DatPySci/pythia-410m-sft-full"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "DatPySci/pythia-410m-sft-full",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/DatPySci/pythia-410m-sft-full

SGLang

How to use DatPySci/pythia-410m-sft-full 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 "DatPySci/pythia-410m-sft-full" \
    --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": "DatPySci/pythia-410m-sft-full",
		"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 "DatPySci/pythia-410m-sft-full" \
        --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": "DatPySci/pythia-410m-sft-full",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use DatPySci/pythia-410m-sft-full with Docker Model Runner:
```
docker model run hf.co/DatPySci/pythia-410m-sft-full
```

DatPySci commited on Apr 20, 2024

Commit

571a3a4

verified ·

1 Parent(s): 1b7574f

End of training

Browse files

Files changed (4) hide show

README.md +6 -2
all_results.json +5 -0
config.json +30 -0
eval_results.json +8 -0

README.md CHANGED Viewed

@@ -2,11 +2,15 @@
 license: apache-2.0
 base_model: EleutherAI/pythia-410m-deduped-v0
 tags:
 - trl
 - sft
 - generated_from_trainer
 datasets:
-- generator
 model-index:
 - name: pythia-410m-sft-full
   results: []
@@ -17,7 +21,7 @@ should probably proofread and complete it, then remove this comment. -->
 # pythia-410m-sft-full
-This model is a fine-tuned version of [EleutherAI/pythia-410m-deduped-v0](https://huggingface.co/EleutherAI/pythia-410m-deduped-v0) on the generator dataset.
 It achieves the following results on the evaluation set:
 - Loss: 1.6286

 license: apache-2.0
 base_model: EleutherAI/pythia-410m-deduped-v0
 tags:
+- alignment-handbook
+- trl
+- sft
+- generated_from_trainer
 - trl
 - sft
 - generated_from_trainer
 datasets:
+- HuggingFaceH4/ultrachat_200k
 model-index:
 - name: pythia-410m-sft-full
   results: []
 # pythia-410m-sft-full
+This model is a fine-tuned version of [EleutherAI/pythia-410m-deduped-v0](https://huggingface.co/EleutherAI/pythia-410m-deduped-v0) on the HuggingFaceH4/ultrachat_200k dataset.
 It achieves the following results on the evaluation set:
 - Loss: 1.6286

all_results.json CHANGED Viewed

@@ -1,5 +1,10 @@
 {
     "epoch": 1.0,
     "train_loss": 1.6493343958478666,
     "train_runtime": 16072.2247,
     "train_samples": 207865,

 {
     "epoch": 1.0,
+    "eval_loss": 1.628558874130249,
+    "eval_runtime": 657.7588,
+    "eval_samples": 23110,
+    "eval_samples_per_second": 21.435,
+    "eval_steps_per_second": 1.341,
     "train_loss": 1.6493343958478666,
     "train_runtime": 16072.2247,
     "train_samples": 207865,

config.json ADDED Viewed

	@@ -0,0 +1,30 @@

+{
+  "_name_or_path": "EleutherAI/pythia-410m-deduped-v0",
+  "architectures": [
+    "GPTNeoXForCausalLM"
+  ],
+  "attention_bias": true,
+  "attention_dropout": 0.0,
+  "bos_token_id": 0,
+  "classifier_dropout": 0.1,
+  "eos_token_id": 0,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.0,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-05,
+  "max_position_embeddings": 2048,
+  "model_type": "gpt_neox",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 24,
+  "rope_scaling": null,
+  "rotary_emb_base": 10000,
+  "rotary_pct": 0.25,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.39.3",
+  "use_cache": true,
+  "use_parallel_residual": true,
+  "vocab_size": 50304
+}

eval_results.json ADDED Viewed

	@@ -0,0 +1,8 @@

+{
+    "epoch": 1.0,
+    "eval_loss": 1.628558874130249,
+    "eval_runtime": 657.7588,
+    "eval_samples": 23110,
+    "eval_samples_per_second": 21.435,
+    "eval_steps_per_second": 1.341
+}