Instructions to use chrlu/zephyr-7b-gemma-hinge with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use chrlu/zephyr-7b-gemma-hinge with Transformers:

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

pipe = pipeline("text-generation", model="chrlu/zephyr-7b-gemma-hinge")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("chrlu/zephyr-7b-gemma-hinge")
model = AutoModelForCausalLM.from_pretrained("chrlu/zephyr-7b-gemma-hinge")
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 chrlu/zephyr-7b-gemma-hinge with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "chrlu/zephyr-7b-gemma-hinge"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "chrlu/zephyr-7b-gemma-hinge",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/chrlu/zephyr-7b-gemma-hinge

SGLang

How to use chrlu/zephyr-7b-gemma-hinge 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 "chrlu/zephyr-7b-gemma-hinge" \
    --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": "chrlu/zephyr-7b-gemma-hinge",
		"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 "chrlu/zephyr-7b-gemma-hinge" \
        --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": "chrlu/zephyr-7b-gemma-hinge",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use chrlu/zephyr-7b-gemma-hinge with Docker Model Runner:
```
docker model run hf.co/chrlu/zephyr-7b-gemma-hinge
```

chrlu commited on Apr 27, 2024

Commit

724cb0f

verified ·

1 Parent(s): 5c0b558

End of training

Browse files

Files changed (5) hide show

README.md +16 -10
all_results.json +13 -0
config.json +1 -1
eval_results.json +16 -0
runs/Apr27_16-40-45_660111d13776/events.out.tfevents.1714235380.660111d13776.64926.1 +3 -0

README.md CHANGED Viewed

@@ -2,9 +2,15 @@
 license: other
 base_model: HuggingFaceH4/zephyr-7b-gemma-sft-v0.1
 tags:
 - trl
 - dpo
 - generated_from_trainer
 model-index:
 - name: zephyr-7b-gemma-hinge
   results: []
@@ -15,17 +21,17 @@ should probably proofread and complete it, then remove this comment. -->
 # zephyr-7b-gemma-hinge
-This model is a fine-tuned version of [HuggingFaceH4/zephyr-7b-gemma-sft-v0.1](https://huggingface.co/HuggingFaceH4/zephyr-7b-gemma-sft-v0.1) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.5250
-- Rewards/chosen: -2.6386
-- Rewards/rejected: -3.9001
-- Rewards/accuracies: 0.7188
-- Rewards/margins: 1.2616
-- Logps/rejected: -440.0739
-- Logps/chosen: -416.4398
-- Logits/rejected: 95.9739
-- Logits/chosen: 101.8169
 ## Model description

 license: other
 base_model: HuggingFaceH4/zephyr-7b-gemma-sft-v0.1
 tags:
+- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
+- trl
+- dpo
+- generated_from_trainer
+datasets:
+- argilla/dpo-mix-7k
 model-index:
 - name: zephyr-7b-gemma-hinge
   results: []
 # zephyr-7b-gemma-hinge
+This model is a fine-tuned version of [HuggingFaceH4/zephyr-7b-gemma-sft-v0.1](https://huggingface.co/HuggingFaceH4/zephyr-7b-gemma-sft-v0.1) on the argilla/dpo-mix-7k dataset.
 It achieves the following results on the evaluation set:
+- Loss: 0.5273
+- Rewards/chosen: -2.6335
+- Rewards/rejected: -3.8935
+- Rewards/accuracies: 0.7292
+- Rewards/margins: 1.2600
+- Logps/rejected: -439.9419
+- Logps/chosen: -416.3391
+- Logits/rejected: 96.0524
+- Logits/chosen: 101.8806
 ## Model description

all_results.json CHANGED Viewed

@@ -1,5 +1,18 @@
 {
     "epoch": 1.971563981042654,
     "total_flos": 0.0,
     "train_loss": 0.4740314678503917,
     "train_runtime": 2291.7376,

 {
     "epoch": 1.971563981042654,
+    "eval_logits/chosen": 101.8805923461914,
+    "eval_logits/rejected": 96.05239868164062,
+    "eval_logps/chosen": -416.339111328125,
+    "eval_logps/rejected": -439.9419250488281,
+    "eval_loss": 0.5273366570472717,
+    "eval_rewards/accuracies": 0.7291666865348816,
+    "eval_rewards/chosen": -2.633545160293579,
+    "eval_rewards/margins": 1.2600029706954956,
+    "eval_rewards/rejected": -3.893548011779785,
+    "eval_runtime": 119.4302,
+    "eval_samples": 750,
+    "eval_samples_per_second": 6.28,
+    "eval_steps_per_second": 0.201,
     "total_flos": 0.0,
     "train_loss": 0.4740314678503917,
     "train_runtime": 2291.7376,

config.json CHANGED Viewed

@@ -24,6 +24,6 @@
   "rope_theta": 10000.0,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.40.1",
-  "use_cache": false,
   "vocab_size": 256000
 }

   "rope_theta": 10000.0,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.40.1",
+  "use_cache": true,
   "vocab_size": 256000
 }

eval_results.json ADDED Viewed

	@@ -0,0 +1,16 @@

+{
+    "epoch": 1.971563981042654,
+    "eval_logits/chosen": 101.8805923461914,
+    "eval_logits/rejected": 96.05239868164062,
+    "eval_logps/chosen": -416.339111328125,
+    "eval_logps/rejected": -439.9419250488281,
+    "eval_loss": 0.5273366570472717,
+    "eval_rewards/accuracies": 0.7291666865348816,
+    "eval_rewards/chosen": -2.633545160293579,
+    "eval_rewards/margins": 1.2600029706954956,
+    "eval_rewards/rejected": -3.893548011779785,
+    "eval_runtime": 119.4302,
+    "eval_samples": 750,
+    "eval_samples_per_second": 6.28,
+    "eval_steps_per_second": 0.201
+}

runs/Apr27_16-40-45_660111d13776/events.out.tfevents.1714235380.660111d13776.64926.1 ADDED Viewed

	@@ -0,0 +1,3 @@

+version https://git-lfs.github.com/spec/v1
+oid sha256:fc4c27d471cbe44ac67012d3908d4834b0c7903b252e1669e73c263f5322827b
+size 815