tune readme around timings etc

Author: karpathy

README.md

@@ -5,11 +5,11 @@ Have you ever wanted to inference a baby [Llama 2](https://ai.meta.com/llama/) m
 
 <img src="assets/llama_cute.jpg" width="300" height="300">
 
-With this code you can train the Llama 2 LLM architecture from scratch in PyTorch, then save the weights to a raw binary file, then load that into one ~simple 500-line C file ([run.c](run.c)) that inferences the model, simply in fp32 for now. On my cloud Linux devbox a dim 288 6-layer 6-head model (~15M params) inferences at ~100 tok/s in fp32, and about the same on my M1 MacBook Air. I was somewhat pleasantly surprised that one can run reasonably sized models (few ten million params) at highly interactive rates with an approach this simple.
+With the code in this repo you can train the Llama 2 LLM architecture from scratch in PyTorch, then export the weights to a binary file, and load that into one ~simple 500-line C file ([run.c](run.c)) that inferences the model. Hence, this repo is a "fullstack" solution to custom, small LLMs. You might think that you need many billion parameter LLMs to do anything useful, but in fact very small LLMs can have surprisingly strong performance if you make the domain narrow enough. I recommend looking at the [TinyStories](https://huggingface.co/datasets/roneneldan/TinyStories) paper for inspiration.
 
-Please note that this is just a weekend project: I took nanoGPT, tuned it to implement the Llama-2 architecture instead of GPT-2, and the meat of it was writing the C inference engine in [run.c](run.c). As such, this is not really meant to be a production-grade library right now.
+Please note that this started as just a fun weekend project: I took nanoGPT, tuned it to implement the Llama-2 architecture instead of GPT-2, and the meat of it was writing the C inference engine in [run.c](run.c). As such, this is not really meant to be a production-grade library right now.
 
-Hat tip to [llama.cpp](https://github.com/ggerganov/llama.cpp) for inspiring this project. I wanted something super minimal so I chose to hard-code the llama-2 architecture, stick to fp32, and just roll one inference file of pure C with no dependencies.
+Hat tip to the awesome [llama.cpp](https://github.com/ggerganov/llama.cpp) for inspiring this project. I wanted something super minimal so I chose to hard-code the llama-2 architecture, stick to fp32, and just roll one inference file of pure C with no dependencies.
 
 ## feel the magic
 
@@ -26,11 +26,11 @@ gcc -O3 -o run run.c -lm
 ./run out/model.bin
 ```
 
-You'll see the text stream a sample. On my M1 MacBook Air this runs at ~100 tokens/s, not bad for super naive fp32 single-threaded C code. See [performance](#performance) for compile flags that can significantly speed this up. Sample output:
+You'll see the text stream a sample. On my M1 MacBook Air this runs at ~110 tokens/s. See [performance](#performance) or the Makefile for compile flags that can significantly speed this up. Sample output:
 
 *Once upon a time, there was a boy named Timmy. Timmy loved to play sports with his friends. He was very good at throwing and catching balls. One day, Timmy's mom gave him a new shirt to wear to a party. Timmy thought it was impressive and asked his mom to explain what a shirt could be for. "A shirt is like a special suit for a basketball game," his mom said. Timmy was happy to hear that and put on his new shirt. He felt like a soldier going to the army and shouting. From that day on, Timmy wore his new shirt every time he played sports with his friends at the party. Once upon a time, there was a little girl named Lily. She loved to play outside with her friends. One day, Lily and her friend Emma were playing with a ball. Emma threw the ball too hard and it hit Lily's face. Lily felt embarrassed and didn't want to play anymore.
 Emma asked Lily what was wrong, and Lily told her about her memory. Emma told Lily that she was embarrassed because she had thrown the ball too hard. Lily felt bad
-achieved tok/s: 98.746993347843922*
+achieved tok/s: 129.146172*
 
 **Update**: I've now also uploaded a bigger checkpoint. This one is dim 512, 8 layers, 8 heads and context length 1024, a ~44M param Transformer. It trained for 200K iterations batch size 32 on 4XA100 40GB GPUs in ~8 hours. You can use this bigger and more powerful checkpoint like so:
 
@@ -39,7 +39,7 @@ wget https://karpathy.ai/llama2c/model44m.bin -P out44m
 ./run out44m/model44m.bin
 ```
 
-On my MacBook Air compiled with $ gcc -Ofast -o run run.c -lm this ran at ~150 tok/s. Still way too fast! I have to train an even bigger checkpoint... This model samples more coherent and diverse stories:
+This still runs at interactive rates and samples more coherent and diverse stories:
 
 *Once upon a time, there was a little girl named Lily. She loved playing with her toys on top of her bed. One day, she decided to have a tea party with her stuffed animals. She poured some tea into a tiny teapot and put it on top of the teapot. Suddenly, her little brother Max came into the room and wanted to join the tea party too. Lily didn't want to share her tea and she told Max to go away. Max started to cry and Lily felt bad. She decided to yield her tea party to Max and they both shared the teapot. But then, something unexpected happened. The teapot started to shake and wiggle. Lily and Max were scared and didn't know what to do. Suddenly, the teapot started to fly towards the ceiling and landed on the top of the bed. Lily and Max were amazed and they hugged each other. They realized that sharing was much more fun than being selfish. From that day on, they always shared their tea parties and toys.*