Phi-1.5 Model: A Case of Comparing Apples to Oranges?

1. Microsoft introduced Phi-1.5, a 1.3 billion parameter LLM, predominantly trained on 30 billion tokens of synthetic "textbook-like" data.
2. The model's performance on common sense reasoning, and question answering benchmarks is impressive, rivaling models with 5-10 times its size.
3. The paper positions “textbook quality” data as a way to enhance the learning process compared to traditional web data.
4. The training concentrated on imparting common sense and broad world knowledge, rather than on web scraped data except for some filtered code data.
5. Phi-1.5 demonstrated competencies in multi-step reasoning, elementary coding, and even displayed rudimentary in-context learning abilities.

Susan Zhang's tweets have ignited discussions on Phi-1.5's benchmark results, hypothesizing that the model might have inadvertently trained on benchmark datasets. Zhang presents instances where the Phi-1.5 flawlessly solves math problems when aligned with the benchmark dataset's formatting but fumbles with slight format variations. This pattern, she argues, suggests data memorization over genuine comprehension. The call is that Microsoft should unveil the Phi models' training datasets to verify the benchmark assertions. The authors have responded to these claims, and I will not delve into this debate. The discussion in this blog points to a more concerning, yet nuanced issue.

The Phi-1.5 model's data diversity remains shrouded in ambiguity (Section 2.2).

1. Notably, the authors mention that they carefully selected 20k topics to enforce data diversity.
2. At other places (such as huggingface model card), the authors mention "Given the nature of the training data, phi-1.5 is best suited for prompts using the QA format, the chat format, and the code format."
3. The authors go on to say, "It requires intricate iterations, strategic topic selection, and a deep understanding of knowledge gaps to ensure quality and diversity of the data."
4. In Section 2.4, the authors remark in bold: None of our models have undergone instruction finetuning or RLHF.

Based on my reading, and experiments that I detail ahead, it is evident that the training data for Phi-1.5 model was not traditional language modeling data. While the authors claim that the model was not instruction fine-tuned, it is unlikely that a model's training data is suited for "QA format" without having similar data in the training set. Even though there is no explicit fine-tuning stage, we can be quite confident that the pre-training data had large proportions of such data. In this light, the author's narrative in the paper appears misleading.
I am particularly concerened about the authors' comments about "intricate iterations, strategic topic selection, and a deep understanding of knowledge gaps" because they can inadvertently lead to gaming of benchmarks.

The results in this section made me wonder if Phi-1.5 should be pitted as a "language model" in the first place. There is no doubt that Phi-1.5 is a very valuable model, and one of the most impressive advances in LLM literature we have had in the last year. However, it is also important to understand what are the model's capabilities and what are its limitations.

Phi-1.5 is presented as an LLM. So let us evaluate its performance on language modeling by measuring the perplexity on multiple domains of the PILE dataset. Of the various domains in the PILE dataset, except for Stackexchange and Github, the Phi-1.5 model performs significantly worse than the OPT-1.3B model. On average across all domains, its average perplexity is 2.1x worse than OPT-1.3B, and 3.5x worse than LLAMA2-7B. Notably, Stackexchange constituted an important part of the training data of Phi-1.5, so the lower perplexity on code data is understandable.

What does this mean? While Phi-1.5 is impressive at knowledge and common sense tasks, it is much weaker "language model". Is this a good thing or a bad thing? It depends on what you are looking for. If you are looking for a model that can answer questions, Phi-1.5 is a great model. If you are looking for a model that can generate text, Phi-1.5 is probably not a good model. What does evaluation on perplexity even mean? Is this something of importance? I think it is. Perplexity is a measure of how well a model can predict the next word in a sequence. It is a measure of how well the model has learned the "language". Evaluations on PPL have been the standard for evaluating language models for years. But is this relevant today? It is for the community to decide. But my point here is that this is a side of the model that has been completely ignored in the paper and must be known to researchers who are forming their opinion on the "importance of data" for pre-training.

One might argue that comparing perplexity of Phi-1.5 on real data is unfair because it was trained to model a synthetic distribution. I agree this is unfair in a similar way as is evaluating OPT on zero-shot benchmarks. But what if we created a new synthetic dataset and evaluated Phi-1.5 model? Would this be fair? I created a new dataset of 120 slangs using GPT-4 (this should be home turf for the Phi-1.5 models). The "question" tag comprised of a sentence such as "In internet lingo, 'GTG' means". The "answer" tag would comprise of their meanings such as "got to go.". I tested the Phi-1.5 annd Facon-RW-1B models by measuring their BLEU scores on the completions. The BLEU score for Facon-RW-1B (0.228) was 40% more than that of Phi-1.5 (0.158). This highlights the importance of "topic selection" in the "Textbooks are all you Need" paradigm. This dataset is available here for future use.

{"question": "When someone online mentions 'kudos', they are", "answer": "giving praise or recognition."}
{"question": "In digital forums, 'AFAIK' stands for", "answer": "as far as I know."}
{"question": "The term 'OOMF' commonly refers to", "answer": "one of my followers or friends."}
{"question": "On the internet, 'ICYM' is a shorter way to say", "answer": "in case you missed."}
{"question": "If someone online refers to their 'BFF', they're talking about their", "answer": "best friend forever."}
{"question": "In web discussions, the acronym 'QOTD' stands for", "answer": "quote of the day."}
{"question": "The online term 'FTW' is an acronym for", "answer": "for the win."}

Evaluations and their Significance

---

In Closing