import numpy as np
mod = 1000000007
def convolution(a, b):
    n = len(a)
    m = len(b)
    l = n + m - 1
    z = 1 << (l - 1).bit_length()
    ah, al = a >> 15, a & ((1 << 15) - 1)
    bh, bl = b >> 15, b & ((1 << 15) - 1)
    p = np.rint(np.fft.irfft(np.fft.rfft(al, z) * np.fft.rfft(bl, z))[:l]).astype(np.int64)
    r = np.rint(np.fft.irfft(np.fft.rfft(ah, z) * np.fft.rfft(bh, z))[:l]).astype(np.int64)
    q = np.rint(np.fft.irfft(np.fft.rfft(al + ah, z) * np.fft.rfft(bl + bh, z))[:l]).astype(np.int64) - p - r
    return ((p + (q << 15)) + (r << 30)) % mod

#[x^n]P/Qを求める
def poly_coef(Q, P, n):
    while n:
        R = np.empty_like(Q)
        R[::2] = Q[::2]
        R[1::2] = -Q[1::2]
        P = convolution(P, R)
        Q = convolution(Q, R)
        P = P[n & 1::2]
        Q = Q[::2]
        n >>= 1
    return P[0]


import sys
input = sys.stdin.readline
k = int(input())
n = int(input())
X  = np.array(list(map(int, input().split())))
Q = np.zeros(k+1, dtype=np.int64)
Q[0] = 1
Q[X] = -1
P = np.array([1])
ans = poly_coef(Q, P, k)
print(ans)