| import itertools | |
| # 1, 3, 25, 443, 13956 | |
| def generate_graphs(n): | |
| """ Generate all directed graphs for n vertices """ | |
| all_possible_edges = [(i, j) for i in range(n) for j in range(n) if i != j] | |
| for edges in itertools.product([False, True], repeat=len(all_possible_edges)): | |
| graph = {i: [] for i in range(n)} | |
| for edge, edge_exists in zip(all_possible_edges, edges): | |
| if edge_exists: |
| SHANA - trnscript cells %@ - index path %@ | |
| sms://open?groupid=%@ | |
| Still presenting %@ | |
| Search "%@" number of results | |
| After clearing "%@" should have no results | |
| %@ -- %@ | |
| %@Units | |
| Failed to enter conversation with chat identifier %@ | |
| Failed to find chat: %@ | |
| %@-%@ |
| .PHONY: create-keypair launch-instance connect-instance shutdown-instance clean log-details | |
| # Variables | |
| KEY_NAME = MyKeyPair | |
| INSTANCE_TYPE = t2.micro | |
| AMI_ID = ami-12345678 | |
| SECURITY_GROUP_ID = sg-12345678 | |
| SUBNET_ID = subnet-12345678 | |
| KEY_FILE = $(KEY_NAME).pem | |
| INSTANCE_IP = # Put your instance's public IP or DNS here |
LangChain's built in parser uses dumb regular expressions.
Simon Willson seems to have hit up against this and looked into tree-sitter.
Berger's OSDI paper at USENIX 2023 which won a best paper award, Triangulating Python Performance Issues with SCALENE seems to be using the profiler's line number ranges to drive queries to ChatGPT.
codreview.gpt part1 - part2 sends several prompts with the last prompt asking for a code review summary.
| name: CICD | |
| on: [push, pull_request] | |
| jobs: | |
| build: | |
| name: Build | |
| runs-on: ${{ matrix.job.os }} | |
| strategy: |
| # ChatGPT4 spit it out with some coaxing - looks legit, haven't tested. | |
| import math | |
| def is_prime(n): | |
| def is_power(a, b): | |
| if a == 1: | |
| return b != 1 | |
| power = 1 | |
| while power < b: |
| # Roman Numeral Evaluation is a Monoid | |
| # Chad Brewbaker | Software Engineer | Telligen.org | |
| # crb002@gmail.com | |
| # Initial release January 15, 2013 | |
| class RomanMonoid | |
| attr_accessor :prefix, :prefix_size, :suffix, :suffix_size, :suffix_credit, :sum, :homo | |
| def initialize(val) |
Asking GPT3.5 to refactor gpt2() into a parallel prefix over inputs with code from:
https://github.com/jaymody/picoGPT/blob/3b7f4d180bb125a76b2f4f7b3a74268e5ec5f131/gpt2_pico.py#L37
def gpt2(inputs, wte, wpe, blocks, ln_f, n_head, chunk_size):
n_chunks = len(inputs) // chunk_size
chunked_inputs = np.split(inputs, n_chunks)
embeddings = []
ChatGPT prompt: How would you write an LLVM optimization pass to lint for register spilling?
An LLVM optimization pass is a way to analyze or transform LLVM Intermediate Representation (IR) code. To write an LLVM optimization pass that lints for register spilling, you would need to analyze the register allocation process and identify cases where register spilling occurs. Here's a high-level outline of the steps you'd need to take:
Set up an LLVM pass: First, you need to set up a new LLVM pass. To do this, create a new class that inherits from llvm::FunctionPass. In this class, you'll implement the runOnFunction method, which will be called for each function in the module you're analyzing.
#include "llvm/Pass.h"
#include "llvm/IR/Function.h"
Guest Thomas Dullien
https://twitter.com/halvarflake