task stringclasses 1
value | input stringlengths 259 305 | output stringclasses 5
values | type stringclasses 1
value | focus stringclasses 1
value | difficulty stringclasses 1
value |
|---|---|---|---|---|---|
coding | Coding Hard 665: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 1850: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 583: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 39: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 1814: Implement RHT Randomized Hadamard Transform with correct case RHT, outlier flattening 10->5.07 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def RHT_Transform(data, signs):
n=len(data)
# Random sign diagonal D
for i in range(n): data[i] *= signs[i] # +/-1
# Hadamard H_n = [H_{n-1} H_{n-1}; H_{n-1} -H_{n-1}]/sqrt(2)
h=1
while h<n:
for i in range(0,n,h*2):
for j in range(i,i+h):
u=data[j]; ... | agentic | coding | hard |
coding | Coding Hard 859: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 1605: Implement NoeSA-24 pack 13 values into 60-bit with base-24 encoding, correct case NoeSA — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def NoeSA_Pack_13x24(values): # NoeSA-24 4.58-bit
assert len(values)==13
packed=0
for i,v in enumerate(values): # v in 0..23
packed += v * (24**i)
# 24^13 ~ 3e17 fits in 60-bit (1.15e18)
return packed.to_bytes(8,'little') # 60-bit in 8 bytes, 4.615 bits/param
Explanation: Implement... | agentic | coding | hard |
coding | Coding Hard 740: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 991: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 585: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 584: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 1056: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 344: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 1072: Implement RHT Randomized Hadamard Transform with correct case RHT, outlier flattening 10->5.07 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def RHT_Transform(data, signs):
n=len(data)
# Random sign diagonal D
for i in range(n): data[i] *= signs[i] # +/-1
# Hadamard H_n = [H_{n-1} H_{n-1}; H_{n-1} -H_{n-1}]/sqrt(2)
h=1
while h<n:
for i in range(0,n,h*2):
for j in range(i,i+h):
u=data[j]; ... | agentic | coding | hard |
coding | Coding Hard 1706: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 748: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 1781: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 599: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 1182: Implement NoeSA-24 pack 13 values into 60-bit with base-24 encoding, correct case NoeSA — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def NoeSA_Pack_13x24(values): # NoeSA-24 4.58-bit
assert len(values)==13
packed=0
for i,v in enumerate(values): # v in 0..23
packed += v * (24**i)
# 24^13 ~ 3e17 fits in 60-bit (1.15e18)
return packed.to_bytes(8,'little') # 60-bit in 8 bytes, 4.615 bits/param
Explanation: Implement... | agentic | coding | hard |
coding | Coding Hard 1818: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 1784: Implement NoeSA-24 pack 13 values into 60-bit with base-24 encoding, correct case NoeSA — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def NoeSA_Pack_13x24(values): # NoeSA-24 4.58-bit
assert len(values)==13
packed=0
for i,v in enumerate(values): # v in 0..23
packed += v * (24**i)
# 24^13 ~ 3e17 fits in 60-bit (1.15e18)
return packed.to_bytes(8,'little') # 60-bit in 8 bytes, 4.615 bits/param
Explanation: Implement... | agentic | coding | hard |
coding | Coding Hard 1618: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 1803: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 1859: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 1826: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 1165: Implement RHT Randomized Hadamard Transform with correct case RHT, outlier flattening 10->5.07 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def RHT_Transform(data, signs):
n=len(data)
# Random sign diagonal D
for i in range(n): data[i] *= signs[i] # +/-1
# Hadamard H_n = [H_{n-1} H_{n-1}; H_{n-1} -H_{n-1}]/sqrt(2)
h=1
while h<n:
for i in range(0,n,h*2):
for j in range(i,i+h):
u=data[j]; ... | agentic | coding | hard |
coding | Coding Hard 1178: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 480: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 1321: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 63: Implement NoeSA-24 pack 13 values into 60-bit with base-24 encoding, correct case NoeSA — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def NoeSA_Pack_13x24(values): # NoeSA-24 4.58-bit
assert len(values)==13
packed=0
for i,v in enumerate(values): # v in 0..23
packed += v * (24**i)
# 24^13 ~ 3e17 fits in 60-bit (1.15e18)
return packed.to_bytes(8,'little') # 60-bit in 8 bytes, 4.615 bits/param
Explanation: Implement... | agentic | coding | hard |
coding | Coding Hard 1066: Implement NoeSA-24 pack 13 values into 60-bit with base-24 encoding, correct case NoeSA — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def NoeSA_Pack_13x24(values): # NoeSA-24 4.58-bit
assert len(values)==13
packed=0
for i,v in enumerate(values): # v in 0..23
packed += v * (24**i)
# 24^13 ~ 3e17 fits in 60-bit (1.15e18)
return packed.to_bytes(8,'little') # 60-bit in 8 bytes, 4.615 bits/param
Explanation: Implement... | agentic | coding | hard |
coding | Coding Hard 1978: Implement RHT Randomized Hadamard Transform with correct case RHT, outlier flattening 10->5.07 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def RHT_Transform(data, signs):
n=len(data)
# Random sign diagonal D
for i in range(n): data[i] *= signs[i] # +/-1
# Hadamard H_n = [H_{n-1} H_{n-1}; H_{n-1} -H_{n-1}]/sqrt(2)
h=1
while h<n:
for i in range(0,n,h*2):
for j in range(i,i+h):
u=data[j]; ... | agentic | coding | hard |
coding | Coding Hard 1810: Implement NoeSA-24 pack 13 values into 60-bit with base-24 encoding, correct case NoeSA — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def NoeSA_Pack_13x24(values): # NoeSA-24 4.58-bit
assert len(values)==13
packed=0
for i,v in enumerate(values): # v in 0..23
packed += v * (24**i)
# 24^13 ~ 3e17 fits in 60-bit (1.15e18)
return packed.to_bytes(8,'little') # 60-bit in 8 bytes, 4.615 bits/param
Explanation: Implement... | agentic | coding | hard |
coding | Coding Hard 1499: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 20: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 1593: Implement NoeSA-24 pack 13 values into 60-bit with base-24 encoding, correct case NoeSA — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def NoeSA_Pack_13x24(values): # NoeSA-24 4.58-bit
assert len(values)==13
packed=0
for i,v in enumerate(values): # v in 0..23
packed += v * (24**i)
# 24^13 ~ 3e17 fits in 60-bit (1.15e18)
return packed.to_bytes(8,'little') # 60-bit in 8 bytes, 4.615 bits/param
Explanation: Implement... | agentic | coding | hard |
coding | Coding Hard 812: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 988: Implement RHT Randomized Hadamard Transform with correct case RHT, outlier flattening 10->5.07 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def RHT_Transform(data, signs):
n=len(data)
# Random sign diagonal D
for i in range(n): data[i] *= signs[i] # +/-1
# Hadamard H_n = [H_{n-1} H_{n-1}; H_{n-1} -H_{n-1}]/sqrt(2)
h=1
while h<n:
for i in range(0,n,h*2):
for j in range(i,i+h):
u=data[j]; ... | agentic | coding | hard |
coding | Coding Hard 877: Implement RHT Randomized Hadamard Transform with correct case RHT, outlier flattening 10->5.07 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def RHT_Transform(data, signs):
n=len(data)
# Random sign diagonal D
for i in range(n): data[i] *= signs[i] # +/-1
# Hadamard H_n = [H_{n-1} H_{n-1}; H_{n-1} -H_{n-1}]/sqrt(2)
h=1
while h<n:
for i in range(0,n,h*2):
for j in range(i,i+h):
u=data[j]; ... | agentic | coding | hard |
coding | Coding Hard 1418: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 779: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 1628: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 744: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 1969: Implement RHT Randomized Hadamard Transform with correct case RHT, outlier flattening 10->5.07 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def RHT_Transform(data, signs):
n=len(data)
# Random sign diagonal D
for i in range(n): data[i] *= signs[i] # +/-1
# Hadamard H_n = [H_{n-1} H_{n-1}; H_{n-1} -H_{n-1}]/sqrt(2)
h=1
while h<n:
for i in range(0,n,h*2):
for j in range(i,i+h):
u=data[j]; ... | agentic | coding | hard |
coding | Coding Hard 1506: Implement NoeSA-24 pack 13 values into 60-bit with base-24 encoding, correct case NoeSA — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def NoeSA_Pack_13x24(values): # NoeSA-24 4.58-bit
assert len(values)==13
packed=0
for i,v in enumerate(values): # v in 0..23
packed += v * (24**i)
# 24^13 ~ 3e17 fits in 60-bit (1.15e18)
return packed.to_bytes(8,'little') # 60-bit in 8 bytes, 4.615 bits/param
Explanation: Implement... | agentic | coding | hard |
coding | Coding Hard 1868: Implement RHT Randomized Hadamard Transform with correct case RHT, outlier flattening 10->5.07 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def RHT_Transform(data, signs):
n=len(data)
# Random sign diagonal D
for i in range(n): data[i] *= signs[i] # +/-1
# Hadamard H_n = [H_{n-1} H_{n-1}; H_{n-1} -H_{n-1}]/sqrt(2)
h=1
while h<n:
for i in range(0,n,h*2):
for j in range(i,i+h):
u=data[j]; ... | agentic | coding | hard |
coding | Coding Hard 285: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 1809: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 897: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 757: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 1642: Implement RHT Randomized Hadamard Transform with correct case RHT, outlier flattening 10->5.07 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def RHT_Transform(data, signs):
n=len(data)
# Random sign diagonal D
for i in range(n): data[i] *= signs[i] # +/-1
# Hadamard H_n = [H_{n-1} H_{n-1}; H_{n-1} -H_{n-1}]/sqrt(2)
h=1
while h<n:
for i in range(0,n,h*2):
for j in range(i,i+h):
u=data[j]; ... | agentic | coding | hard |
coding | Coding Hard 1271: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 886: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 872: Implement RHT Randomized Hadamard Transform with correct case RHT, outlier flattening 10->5.07 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def RHT_Transform(data, signs):
n=len(data)
# Random sign diagonal D
for i in range(n): data[i] *= signs[i] # +/-1
# Hadamard H_n = [H_{n-1} H_{n-1}; H_{n-1} -H_{n-1}]/sqrt(2)
h=1
while h<n:
for i in range(0,n,h*2):
for j in range(i,i+h):
u=data[j]; ... | agentic | coding | hard |
coding | Coding Hard 791: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 1070: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 751: Implement RHT Randomized Hadamard Transform with correct case RHT, outlier flattening 10->5.07 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def RHT_Transform(data, signs):
n=len(data)
# Random sign diagonal D
for i in range(n): data[i] *= signs[i] # +/-1
# Hadamard H_n = [H_{n-1} H_{n-1}; H_{n-1} -H_{n-1}]/sqrt(2)
h=1
while h<n:
for i in range(0,n,h*2):
for j in range(i,i+h):
u=data[j]; ... | agentic | coding | hard |
coding | Coding Hard 512: Implement NoeSA-24 pack 13 values into 60-bit with base-24 encoding, correct case NoeSA — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def NoeSA_Pack_13x24(values): # NoeSA-24 4.58-bit
assert len(values)==13
packed=0
for i,v in enumerate(values): # v in 0..23
packed += v * (24**i)
# 24^13 ~ 3e17 fits in 60-bit (1.15e18)
return packed.to_bytes(8,'little') # 60-bit in 8 bytes, 4.615 bits/param
Explanation: Implement... | agentic | coding | hard |
coding | Coding Hard 5: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 1433: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 1889: Implement NoeSA-24 pack 13 values into 60-bit with base-24 encoding, correct case NoeSA — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def NoeSA_Pack_13x24(values): # NoeSA-24 4.58-bit
assert len(values)==13
packed=0
for i,v in enumerate(values): # v in 0..23
packed += v * (24**i)
# 24^13 ~ 3e17 fits in 60-bit (1.15e18)
return packed.to_bytes(8,'little') # 60-bit in 8 bytes, 4.615 bits/param
Explanation: Implement... | agentic | coding | hard |
coding | Coding Hard 549: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 520: Implement NoeSA-24 pack 13 values into 60-bit with base-24 encoding, correct case NoeSA — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def NoeSA_Pack_13x24(values): # NoeSA-24 4.58-bit
assert len(values)==13
packed=0
for i,v in enumerate(values): # v in 0..23
packed += v * (24**i)
# 24^13 ~ 3e17 fits in 60-bit (1.15e18)
return packed.to_bytes(8,'little') # 60-bit in 8 bytes, 4.615 bits/param
Explanation: Implement... | agentic | coding | hard |
coding | Coding Hard 1481: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 569: Implement NoeSA-24 pack 13 values into 60-bit with base-24 encoding, correct case NoeSA — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def NoeSA_Pack_13x24(values): # NoeSA-24 4.58-bit
assert len(values)==13
packed=0
for i,v in enumerate(values): # v in 0..23
packed += v * (24**i)
# 24^13 ~ 3e17 fits in 60-bit (1.15e18)
return packed.to_bytes(8,'little') # 60-bit in 8 bytes, 4.615 bits/param
Explanation: Implement... | agentic | coding | hard |
coding | Coding Hard 293: Implement NoeSA-24 pack 13 values into 60-bit with base-24 encoding, correct case NoeSA — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def NoeSA_Pack_13x24(values): # NoeSA-24 4.58-bit
assert len(values)==13
packed=0
for i,v in enumerate(values): # v in 0..23
packed += v * (24**i)
# 24^13 ~ 3e17 fits in 60-bit (1.15e18)
return packed.to_bytes(8,'little') # 60-bit in 8 bytes, 4.615 bits/param
Explanation: Implement... | agentic | coding | hard |
coding | Coding Hard 843: Implement RHT Randomized Hadamard Transform with correct case RHT, outlier flattening 10->5.07 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def RHT_Transform(data, signs):
n=len(data)
# Random sign diagonal D
for i in range(n): data[i] *= signs[i] # +/-1
# Hadamard H_n = [H_{n-1} H_{n-1}; H_{n-1} -H_{n-1}]/sqrt(2)
h=1
while h<n:
for i in range(0,n,h*2):
for j in range(i,i+h):
u=data[j]; ... | agentic | coding | hard |
coding | Coding Hard 572: Implement RHT Randomized Hadamard Transform with correct case RHT, outlier flattening 10->5.07 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def RHT_Transform(data, signs):
n=len(data)
# Random sign diagonal D
for i in range(n): data[i] *= signs[i] # +/-1
# Hadamard H_n = [H_{n-1} H_{n-1}; H_{n-1} -H_{n-1}]/sqrt(2)
h=1
while h<n:
for i in range(0,n,h*2):
for j in range(i,i+h):
u=data[j]; ... | agentic | coding | hard |
coding | Coding Hard 173: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 719: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 1619: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 678: Implement RHT Randomized Hadamard Transform with correct case RHT, outlier flattening 10->5.07 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def RHT_Transform(data, signs):
n=len(data)
# Random sign diagonal D
for i in range(n): data[i] *= signs[i] # +/-1
# Hadamard H_n = [H_{n-1} H_{n-1}; H_{n-1} -H_{n-1}]/sqrt(2)
h=1
while h<n:
for i in range(0,n,h*2):
for j in range(i,i+h):
u=data[j]; ... | agentic | coding | hard |
coding | Coding Hard 220: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 631: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 573: Implement NoeSA-24 pack 13 values into 60-bit with base-24 encoding, correct case NoeSA — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def NoeSA_Pack_13x24(values): # NoeSA-24 4.58-bit
assert len(values)==13
packed=0
for i,v in enumerate(values): # v in 0..23
packed += v * (24**i)
# 24^13 ~ 3e17 fits in 60-bit (1.15e18)
return packed.to_bytes(8,'little') # 60-bit in 8 bytes, 4.615 bits/param
Explanation: Implement... | agentic | coding | hard |
coding | Coding Hard 195: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 876: Implement RHT Randomized Hadamard Transform with correct case RHT, outlier flattening 10->5.07 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def RHT_Transform(data, signs):
n=len(data)
# Random sign diagonal D
for i in range(n): data[i] *= signs[i] # +/-1
# Hadamard H_n = [H_{n-1} H_{n-1}; H_{n-1} -H_{n-1}]/sqrt(2)
h=1
while h<n:
for i in range(0,n,h*2):
for j in range(i,i+h):
u=data[j]; ... | agentic | coding | hard |
coding | Coding Hard 651: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 464: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 98: Implement RHT Randomized Hadamard Transform with correct case RHT, outlier flattening 10->5.07 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def RHT_Transform(data, signs):
n=len(data)
# Random sign diagonal D
for i in range(n): data[i] *= signs[i] # +/-1
# Hadamard H_n = [H_{n-1} H_{n-1}; H_{n-1} -H_{n-1}]/sqrt(2)
h=1
while h<n:
for i in range(0,n,h*2):
for j in range(i,i+h):
u=data[j]; ... | agentic | coding | hard |
coding | Coding Hard 1933: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 383: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 319: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 1170: Implement NoeSA-24 pack 13 values into 60-bit with base-24 encoding, correct case NoeSA — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def NoeSA_Pack_13x24(values): # NoeSA-24 4.58-bit
assert len(values)==13
packed=0
for i,v in enumerate(values): # v in 0..23
packed += v * (24**i)
# 24^13 ~ 3e17 fits in 60-bit (1.15e18)
return packed.to_bytes(8,'little') # 60-bit in 8 bytes, 4.615 bits/param
Explanation: Implement... | agentic | coding | hard |
coding | Coding Hard 1393: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 1105: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 1218: Implement RHT Randomized Hadamard Transform with correct case RHT, outlier flattening 10->5.07 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def RHT_Transform(data, signs):
n=len(data)
# Random sign diagonal D
for i in range(n): data[i] *= signs[i] # +/-1
# Hadamard H_n = [H_{n-1} H_{n-1}; H_{n-1} -H_{n-1}]/sqrt(2)
h=1
while h<n:
for i in range(0,n,h*2):
for j in range(i,i+h):
u=data[j]; ... | agentic | coding | hard |
coding | Coding Hard 901: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 136: Implement NoeSA-24 pack 13 values into 60-bit with base-24 encoding, correct case NoeSA — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def NoeSA_Pack_13x24(values): # NoeSA-24 4.58-bit
assert len(values)==13
packed=0
for i,v in enumerate(values): # v in 0..23
packed += v * (24**i)
# 24^13 ~ 3e17 fits in 60-bit (1.15e18)
return packed.to_bytes(8,'little') # 60-bit in 8 bytes, 4.615 bits/param
Explanation: Implement... | agentic | coding | hard |
coding | Coding Hard 1510: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 130: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 1338: Implement NoeSA-24 pack 13 values into 60-bit with base-24 encoding, correct case NoeSA — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def NoeSA_Pack_13x24(values): # NoeSA-24 4.58-bit
assert len(values)==13
packed=0
for i,v in enumerate(values): # v in 0..23
packed += v * (24**i)
# 24^13 ~ 3e17 fits in 60-bit (1.15e18)
return packed.to_bytes(8,'little') # 60-bit in 8 bytes, 4.615 bits/param
Explanation: Implement... | agentic | coding | hard |
coding | Coding Hard 806: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 842: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 1288: Implement Ntarra-DnA phase rotator with correct case Ntarra-DnA, Direction NEG/ZERO/POS, Phase PHI0/1/2 shifts 0/2/4 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def Ntarra_DnA_Compute(act, dir, phase): # Ntarra-DnA 3.17-bit
shift = {0:0, 1:2, 2:4}[phase] # PHI0 0, PHI1 2, PHI2 4
shifted = act << shift
if dir == -1: return -shifted # NEG
elif dir == 0: return 0 # ZERO
else: return shifted # POS
# Pack 2x9: 9*9=81 combos in 7-bit, 47 codes lefto... | agentic | coding | hard |
coding | Coding Hard 103: Implement RHT Randomized Hadamard Transform with correct case RHT, outlier flattening 10->5.07 — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def RHT_Transform(data, signs):
n=len(data)
# Random sign diagonal D
for i in range(n): data[i] *= signs[i] # +/-1
# Hadamard H_n = [H_{n-1} H_{n-1}; H_{n-1} -H_{n-1}]/sqrt(2)
h=1
while h<n:
for i in range(0,n,h*2):
for j in range(i,i+h):
u=data[j]; ... | agentic | coding | hard |
coding | Coding Hard 1443: Implement SatU1 AVX2 kernel with correct case SatU1, not SATU1, for Ryzen5 5650U — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: inline __m256i SatU1_XNOR_AVX2(__m256i a, __m256i b) { __m256i xor_res = _mm256_xor_si256(a,b); return _mm256_andnot_si256(xor_res, _mm256_set1_epi32(-1)); } // XNOR = NOT XOR, then popcount via _mm256_sad_epu8 for 16 Giga/s
Explanation: Implements IKNN-Rl1-A1 tri-tier kernel for CPU-first LLM agentic researc... | agentic | coding | hard |
coding | Coding Hard 1806: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 472: Implement PG-KVC Phase-Gated KV Cache 94% saving with correct case PG-KVC — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def PG_KVC_Push(k,v,entropy):
prec = 1 if entropy<0.5 else 2 # Phase-gated: low entropy 1-bit, high 2-bit
orig = len(k)*4 + len(v)*4
comp = (len(k)+len(v))*prec//8
saving = 1 - comp/orig # 94% saving target -80% PASS
return comp, orig, saving
Explanation: Implements IKNN-Rl1-A1 tri-tier ke... | agentic | coding | hard |
coding | Coding Hard 442: Implement NoeSA-24 pack 13 values into 60-bit with base-24 encoding, correct case NoeSA — repo deeprcurs/IKNN-Rl1-A1 — model IKNN-Rl1-A1 — file IKNN-Rl1-A1-150M.iknn at root beside iknn.cpp — tidy — correct case SatU1 NoeSA Ntarra-DnA MoE RHT PG-KVC PEP ADLP | Solution: def NoeSA_Pack_13x24(values): # NoeSA-24 4.58-bit
assert len(values)==13
packed=0
for i,v in enumerate(values): # v in 0..23
packed += v * (24**i)
# 24^13 ~ 3e17 fits in 60-bit (1.15e18)
return packed.to_bytes(8,'little') # 60-bit in 8 bytes, 4.615 bits/param
Explanation: Implement... | agentic | coding | hard |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.