YLGH

b=128, s_q=1, mean_sk=4096, h_q=16, h_kv=1, d=576, dv=512, causal=True, varlen=False
0.164 ms, 112 TFLOPS, 3718 GB/s
b=128, s_q=1, mean_sk=4096, h_q=16, h_kv=1, d=576, dv=512, causal=True, varlen=True
0.175 ms, 105 TFLOPS, 3498 GB/s
b=128, s_q=2, mean_sk=4096, h_q=16, h_kv=1, d=576, dv=512, causal=True, varlen=False
0.168 ms, 218 TFLOPS, 3659 GB/s
b=128, s_q=2, mean_sk=4096, h_q=16, h_kv=1, d=576, dv=512, causal=True, varlen=True
0.184 ms, 205 TFLOPS, 3445 GB/s
b=128, s_q=1, mean_sk=4096, h_q=32, h_kv=1, d=576, dv=512, causal=True, varlen=False
0.166 ms, 220 TFLOPS, 3685 GB/s

YLGH

#0 building with "default" instance using docker driver

#1 [internal] load .dockerignore
#1 transferring context: 2B done
#1 ...

#2 [internal] load build definition from Dockerfile.cuda
#2 transferring dockerfile: 7.37kB done
#2 DONE 0.3s

YLGH

2024-11-26 05:18:13,602 INFO prefiller:0-0:3853225:140670816544576 prefill_worker.py:886] Prefill worker step took more than 3 seconds, FIND ME YING 10.254240989685059
2024-11-26 05:18:43,639 INFO prefiller:0-0:3853225:140670816544576 prefill_worker.py:886] Prefill worker step took more than 3 seconds, FIND ME YING 9.305031538009644
2024-11-26 05:19:13,677 INFO prefiller:0-0:3853225:140670816544576 prefill_worker.py:886] Prefill worker step took more than 3 seconds, FIND ME YING 9.307690858840942
2024-11-26 05:19:43,698 INFO prefiller:0-0:3853225:140670816544576 prefill_worker.py:886] Prefill worker step took more than 3 seconds, FIND ME YING 9.345906019210815
2024-11-26 05:20:13,719 INFO prefiller:0-0:3853225:140670816544576 prefill_worker.py:886] Prefill worker step took more than 3 seconds, FIND ME YING 9.400884866714478
2024-11-26 05:20:43,751 INFO prefiller:0-0:3853225:140670816544576 prefill_worker.py:886] Prefill worker step took more than 3 seconds, FIND ME YING 9.450013399124146
2024-11-26 05:21:13,7

YLGH

#include <vector>
#include <iostream>
#include <algorithm>
#include <map>

using namespace std;

int T;
int N;
int K;

YLGH

class Solution:
    def minNumberOfFrogs(self, croakOfFrogs: str) -> int:
        index = {}
        for x in 'croak':
            index[x] = []
        for i,c in enumerate(croakOfFrogs):
            index[c].append(i)
        def foo():
            ans = []                           
            for c_index in index['c']:

YLGH

class Solution:
    def wordBreak(self, S, wordDict):
        words = set(wordDict)
        memo = {}
        def dp(index):
            if index == len(S):
                return [""]
            if index in memo:
                return memo[index]

YLGH

from collections import deque
def minKBitFlips(A,K):
    cardinality = 0
    count = 0
    flip = [0 for _ in range(len(A) + 1)]
    for i in range(len(A)):
        cardinality += flip[i]
        if (A[i] + cardinality) % 2 == 1:
            continue
        else:

YLGH

class Solution:
    def minKBitFlips(self, A: List[int], K: int) -> int:
        cardinality = 0
        count = 0
        flip = [0 for _ in range(len(A)+1)]
        for i in range(len(A)):
            cardinality += flip[i]      
            if (A[i] + cardinality) % 2 == 1:
                continue
            else:

YLGH

import heapq
class Solution:
    def maxPerformance(self, N, speed, efficiency, K):
        workers = sorted(list(zip(speed, efficiency)) ,key=lambda x: x[1])
        sm = 0
        ans = 0
        hp = []
        for i in range(len(workers)-1,-1,-1):
            worker = workers[i]
            if len(hp) < K:

YLGH

class Solution:
    def shortestCommonSupersequence(self, str1: str, str2: str) -> str:
        str1 = [None]+[x for x in str1]
        str2 = [None]+[x for x in str2]

        M = len(str1)
        N = len(str2)
        # len, last char, pointer to parent
        dp = [[ (None,None,None) for _ in range(N)] for _ in range(M)]