Instructions to use amdevraj/mistral-7b-ift with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use amdevraj/mistral-7b-ift with Transformers:

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

pipe = pipeline("text-generation", model="amdevraj/mistral-7b-ift")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("amdevraj/mistral-7b-ift")
model = AutoModelForCausalLM.from_pretrained("amdevraj/mistral-7b-ift")
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 amdevraj/mistral-7b-ift with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/amdevraj/mistral-7b-ift

SGLang

How to use amdevraj/mistral-7b-ift 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 "amdevraj/mistral-7b-ift" \
    --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": "amdevraj/mistral-7b-ift",
		"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 "amdevraj/mistral-7b-ift" \
        --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": "amdevraj/mistral-7b-ift",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use amdevraj/mistral-7b-ift with Docker Model Runner:
```
docker model run hf.co/amdevraj/mistral-7b-ift
```

amdevraj commited on Oct 24, 2024

Commit

877a9ce

verified ·

1 Parent(s): bd86d6b

Model save

Browse files

Files changed (4) hide show

README.md +11 -15
all_results.json +6 -11
train_results.json +6 -6
trainer_state.json +0 -0

README.md CHANGED Viewed

@@ -3,15 +3,11 @@ library_name: transformers
 license: apache-2.0
 base_model: mistralai/Mistral-7B-v0.3
 tags:
-- alignment-handbook
-- trl
-- sft
-- generated_from_trainer
 - trl
 - sft
 - generated_from_trainer
 datasets:
-- data/ift
 model-index:
 - name: mistral-7b-ift
   results: []
@@ -22,9 +18,9 @@ should probably proofread and complete it, then remove this comment. -->
 # mistral-7b-ift
-This model is a fine-tuned version of [mistralai/Mistral-7B-v0.3](https://huggingface.co/mistralai/Mistral-7B-v0.3) on the data/ift dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.9144
 ## Model description
@@ -48,10 +44,10 @@ The following hyperparameters were used during training:
 - eval_batch_size: 8
 - seed: 42
 - distributed_type: multi-GPU
-- num_devices: 4
 - gradient_accumulation_steps: 4
-- total_train_batch_size: 128
-- total_eval_batch_size: 32
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine
 - lr_scheduler_warmup_ratio: 0.1
@@ -59,11 +55,11 @@ The following hyperparameters were used during training:
 ### Training results
-| Training Loss | Epoch | Step | Validation Loss |
-|:-------------:|:-----:|:----:|:---------------:|
-| 0.9502        | 1.0   | 911  | 0.9853          |
-| 0.8625        | 2.0   | 1822 | 0.9243          |
-| 0.8581        | 3.0   | 2733 | 0.9144          |
 ### Framework versions

 license: apache-2.0
 base_model: mistralai/Mistral-7B-v0.3
 tags:
 - trl
 - sft
 - generated_from_trainer
 datasets:
+- generator
 model-index:
 - name: mistral-7b-ift
   results: []
 # mistral-7b-ift
+This model is a fine-tuned version of [mistralai/Mistral-7B-v0.3](https://huggingface.co/mistralai/Mistral-7B-v0.3) on the generator dataset.
 It achieves the following results on the evaluation set:
+- Loss: 0.9529
 ## Model description
 - eval_batch_size: 8
 - seed: 42
 - distributed_type: multi-GPU
+- num_devices: 8
 - gradient_accumulation_steps: 4
+- total_train_batch_size: 256
+- total_eval_batch_size: 64
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine
 - lr_scheduler_warmup_ratio: 0.1
 ### Training results
+| Training Loss | Epoch  | Step | Validation Loss |
+|:-------------:|:------:|:----:|:---------------:|
+| 0.9643        | 0.9989 | 455  | 1.0198          |
+| 0.9077        | 2.0    | 911  | 0.9618          |
+| 0.8919        | 2.9967 | 1365 | 0.9529          |
 ### Framework versions

all_results.json CHANGED Viewed

@@ -1,14 +1,9 @@
 {
-    "epoch": 3.0,
-    "eval_loss": 0.914355993270874,
-    "eval_runtime": 6.1247,
-    "eval_samples": 512,
-    "eval_samples_per_second": 43.758,
-    "eval_steps_per_second": 1.469,
-    "total_flos": 572182254059520.0,
-    "train_loss": 0.9187763968187508,
-    "train_runtime": 32723.3769,
     "train_samples": 77486,
-    "train_samples_per_second": 10.687,
-    "train_steps_per_second": 0.084
 }

 {
+    "epoch": 2.996706915477497,
+    "total_flos": 571501770178560.0,
+    "train_loss": 0.9509169136648213,
+    "train_runtime": 15780.8295,
     "train_samples": 77486,
+    "train_samples_per_second": 22.162,
+    "train_steps_per_second": 0.086
 }

train_results.json CHANGED Viewed

@@ -1,9 +1,9 @@
 {
-    "epoch": 3.0,
-    "total_flos": 572182254059520.0,
-    "train_loss": 0.9187763968187508,
-    "train_runtime": 32723.3769,
     "train_samples": 77486,
-    "train_samples_per_second": 10.687,
-    "train_steps_per_second": 0.084
 }

 {
+    "epoch": 2.996706915477497,
+    "total_flos": 571501770178560.0,
+    "train_loss": 0.9509169136648213,
+    "train_runtime": 15780.8295,
     "train_samples": 77486,
+    "train_samples_per_second": 22.162,
+    "train_steps_per_second": 0.086
 }

trainer_state.json CHANGED Viewed

The diff for this file is too large to render. See raw diff