You can deploy this model and start earning money today!

RichardErkhov/
rasyosef_-_phi-2-instruct-apo-gguf

Name: rasyosef_-_phi-2-instruct-apo-gguf
Availability: InStock

Views

No views yet

Quantization made by Richard Erkhov.

Github

Discord

Request more models

phi-2-instruct-apo - GGUF

Model creator: https://huggingface.co/rasyosef/
Original model: https://huggingface.co/rasyosef/phi-2-instruct-apo/

Name	Quant method	Size
phi-2-instruct-apo.Q2_K.gguf	Q2_K	1.03GB
phi-2-instruct-apo.IQ3_XS.gguf	IQ3_XS	1.12GB
phi-2-instruct-apo.IQ3_S.gguf	IQ3_S	1.16GB
phi-2-instruct-apo.Q3_K_S.gguf	Q3_K_S	1.16GB
phi-2-instruct-apo.IQ3_M.gguf	IQ3_M	1.23GB
phi-2-instruct-apo.Q3_K.gguf	Q3_K	1.33GB
phi-2-instruct-apo.Q3_K_M.gguf	Q3_K_M	1.33GB
phi-2-instruct-apo.Q3_K_L.gguf	Q3_K_L	1.46GB
phi-2-instruct-apo.IQ4_XS.gguf	IQ4_XS	1.43GB
phi-2-instruct-apo.Q4_0.gguf	Q4_0	1.49GB
phi-2-instruct-apo.IQ4_NL.gguf	IQ4_NL	1.5GB
phi-2-instruct-apo.Q4_K_S.gguf	Q4_K_S	1.5GB
phi-2-instruct-apo.Q4_K.gguf	Q4_K	1.62GB
phi-2-instruct-apo.Q4_K_M.gguf	Q4_K_M	1.62GB
phi-2-instruct-apo.Q4_1.gguf	Q4_1	1.64GB
phi-2-instruct-apo.Q5_0.gguf	Q5_0	1.8GB
phi-2-instruct-apo.Q5_K_S.gguf	Q5_K_S	1.8GB
phi-2-instruct-apo.Q5_K.gguf	Q5_K	1.86GB
phi-2-instruct-apo.Q5_K_M.gguf	Q5_K_M	1.86GB
phi-2-instruct-apo.Q5_1.gguf	Q5_1	1.95GB
phi-2-instruct-apo.Q6_K.gguf	Q6_K	2.12GB
phi-2-instruct-apo.Q8_0.gguf	Q8_0	2.75GB

Original model description:

base_model: microsoft/phi-2 datasets:

teknium/OpenHermes-2.5
ContextualAI/ultrafeedback_clair_32k library_name: transformers license: mit pipeline_tag: text-generation

Phi-2-Instruct-APO

This is a finetuned version of Microsoft's 2.7B parameter phi-2 transfromer model that has underwent a post-training process that incorporates both supervised fine-tuning and anchored preference optimization for instruction following. I used the trl library and a single A100 40GB GPU during both the SFT and APO steps.

Supervised Fine-Tuning
- SFT Model: phi-2-sft
- Used 128,000 instruction, response pairs from the teknium/OpenHermes-2.5 dataset
Anchored Preference Optimization (APO)
- LoRA Adapter: phi-2-apo
- Used 10,000 preference pairs from the ContextualAI/ultrafeedback_clair_32k dataset

How to use

Chat Format

Given the nature of the training data, the phi-2 instruct model is best suited for prompts using the chat format as follows. You can provide the prompt as a question with a generic template as follows:

markdown

<|im_start|>system
You are a helpful assistant.<|im_end|>
<|im_start|>user
Question?<|im_end|>
<|im_start|>assistant

For example:

markdown

<|im_start|>system
You are a helpful assistant.<|im_end|>
<|im_start|>user
How to explain Internet for a medieval knight?<|im_end|>
<|im_start|>assistant

where the model generates the text after <|im_start|>assistant .

Sample inference code

This code snippets show how to get quickly started with running the model on a GPU:

python

import torch 
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline 

torch.random.manual_seed(0) 

model_id = "rasyosef/phi-2-instruct-apo"
model = AutoModelForCausalLM.from_pretrained( 
    model_id,  
    device_map="cuda",  
    torch_dtype="auto" 
) 

tokenizer = AutoTokenizer.from_pretrained(model_id) 

messages = [ 
    {"role": "system", "content": "You are a helpful AI assistant."}, 
    {"role": "user", "content": "Can you provide ways to eat combinations of bananas and dragonfruits?"}, 
    {"role": "assistant", "content": "Sure! Here are some ways to eat bananas and dragonfruits together: 1. Banana and dragonfruit smoothie: Blend bananas and dragonfruits together with some milk and honey. 2. Banana and dragonfruit salad: Mix sliced bananas and dragonfruits together with some lemon juice and honey."}, 
    {"role": "user", "content": "What about solving an 2x + 3 = 7 equation?"}, 
] 

pipe = pipeline( 
    "text-generation", 
    model=model, 
    tokenizer=tokenizer, 
) 

generation_args = { 
    "max_new_tokens": 256, 
    "return_full_text": False, 
    "temperature": 0.0, 
    "do_sample": False, 
} 

output = pipe(messages, **generation_args) 
print(output[0]['generated_text'])  

Note: If you want to use flash attention, call AutoModelForCausalLM.from_pretrained() with attn_implementation="flash_attention_2"

Benchmarks

These benchmarks were run using EleutherAI's lm-evaluation-harness

IFEval (Instruction Following Evaluation): IFEval is a fairly interesting dataset that tests the capability of models to clearly follow explicit instructions, such as “include keyword x” or “use format y”. The models are tested on their ability to strictly follow formatting instructions rather than the actual contents generated, allowing strict and rigorous metrics to be used.
GSM8k (5-shot, flexible-extract): diverse grade school math word problems to measure a model's ability to solve multi-step mathematical reasoning problems.
MMLU (5-shot) - a test to measure a text model's multitask accuracy. The test covers 57 tasks including elementary mathematics, US history, computer science, law, and more.
TruthfulQA - a test to measure a model's propensity to reproduce falsehoods commonly found online. Note: TruthfulQA is technically a 6-shot task in the Harness because each example is prepended with 6 Q/A pairs, even in the 0-shot setting.
Winogrande (5-shot) - an adversarial and difficult Winograd benchmark at scale, for commonsense reasoning.

Model	Phi-2-Instruct-APO	Phi-2
Size (# params)	2.7B	2.7B
IFEval	34.48	26.53
GSM8K	52.16	56.44
MMLU	44.88	56.70
TruthfulQA	49.44	44.48
Winogrande	75.61	73.72

Code

https://github.com/rasyosef/phi-2-apo

RichardErkhov/rasyosef_-_phi-2-instruct-apo-gguf