patrickbdevaney commited on
Commit
dfd6ea8
·
verified ·
1 Parent(s): 3da2734

archive arm s2-abl-ce0.9_tv0.1: eval.log

Browse files
Files changed (1) hide show
  1. arms/s2-abl-ce0.9_tv0.1/eval.log +1201 -0
arms/s2-abl-ce0.9_tv0.1/eval.log ADDED
@@ -0,0 +1,1201 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [decode] loading /home/patrickd/models/DeepSeek-V4-Flash-0731-REAP ... s=6 NDEC=8 seqmax=298
2
+ [decode] loaded 100.40 GiB, 45821 tensors (weights in MANAGED device-preferred memory)
3
+ [decode] building 43 layer structs once (persistent)...
4
+ [decode] structs built. mem 108.1/122.8 GiB
5
+ [decode] prefill 5 positions...
6
+ [decode] PREFILL: 5 positions in 230.8 ms = 21.7 tok/s (46.152 ms/tok)
7
+ [decode] prefill done. caches populated. starting decode.
8
+ step 0 pos 5 -> token 11111 (71.0 ms warmup)
9
+ step 1 pos 6 -> token 16 (71.3 ms)
10
+ step 2 pos 7 -> token 455 (78.8 ms)
11
+ step 3 pos 8 -> token 6102 (68.8 ms)
12
+ step 4 pos 9 -> token 294 (67.0 ms)
13
+ step 5 pos 10 -> token 16603 (67.3 ms)
14
+ step 6 pos 11 -> token 344 (75.1 ms)
15
+ step 7 pos 12 -> token 29168 (66.8 ms)
16
+
17
+ [decode] first decoded token argmax = 11111 (expect 11111) -> GATE PASS
18
+ [decode] generated: 11111 16 455 6102 294 16603 344 29168
19
+ [decode] WARM decode: 70.7 ms/tok = 14.14 tok/s (M=1 steady state, 7-step avg)
20
+
21
+ [spec-verify] M=5 verify in ONE forward: 117.9 ms (= 23.6 ms/tok if all accepted vs 70.7 M=1 -> 3.00x)
22
+ [spec-verify] verify argmax: 11111 16 455 6102 294
23
+ [spec-verify] decode tokens: 11111 16 455 6102 294
24
+ [spec-verify] MATCH 5/5 -> PASS (M=K verify == K sequential decodes; diffs = MoE-atomic near-ties)
25
+
26
+ [spec] loading EXTERNAL DSpark head /home/patrickd/s5-capture/s2/abl_ce0.9_tv0.1_head ...
27
+ [spec] head loaded 6.53 GiB, 2977 mtp tensors (SEPARATE store)
28
+ [spec] NSTAGE=3 head-experts=160 BLK sweep: 6 6 6 6 6 6 6 6 6
29
+ [spec] head built. mem 115.6/122.8 GiB
30
+ [decode] PREFILL: 5 positions in 145.6 ms = 34.3 tok/s (29.119 ms/tok)
31
+ [spec] decoding 200 tokens (block=6, draft passes=1, adaptK=1.50, prompt=0 s=6)...
32
+ [arena] first draft: draft footprint 214.32 MB, global hwm 214.32 MB / cap 512.00 MB (41.9%), draft path = ARENA
33
+ [margins] 2.13 0.17 0.07 5.59 1.75 0.14 K=2
34
+ verify 1: accepted 1/1 (K=2) + correction -> +2 tokens (110.0 ms) cpos=7
35
+ [margins] 0.33 1.30 0.83 0.10 0.07 1.55 K=2
36
+ verify 2: accepted 0/1 (K=2) + correction -> +1 tokens (119.4 ms) cpos=8
37
+ [margins] 2.87 1.53 0.16 4.12 2.03 0.19 K=3
38
+ verify 3: accepted 2/2 (K=3) + correction -> +3 tokens (114.8 ms) cpos=11
39
+ [margins] 6.22 0.13 0.46 2.99 2.55 4.15 K=2
40
+ verify 4: accepted 1/1 (K=2) + correction -> +2 tokens (105.9 ms) cpos=13
41
+ [margins] 3.37 4.90 4.86 1.42 0.15 4.77 K=4
42
+ verify 5: accepted 3/3 (K=4) + correction -> +4 tokens (133.7 ms) cpos=17
43
+ [margins] 0.07 3.65 1.14 2.17 2.32 3.86 K=3
44
+ verify 6: accepted 2/2 (K=3) + correction -> +3 tokens (118.4 ms) cpos=20
45
+ [margins] 2.56 3.39 4.73 1.25 0.12 2.57 K=4
46
+ verify 7: accepted 3/3 (K=4) + correction -> +4 tokens (134.1 ms) cpos=24
47
+ [margins] 0.74 6.32 0.63 3.74 1.22 5.80 K=3
48
+ verify 8: accepted 0/2 (K=3) + correction -> +1 tokens (111.0 ms) cpos=25
49
+ [margins] 8.79 1.90 4.63 2.04 4.56 3.44 K=7
50
+ verify 9: accepted 1/6 (K=7) + correction -> +2 tokens (184.6 ms) cpos=27
51
+ [margins] 3.02 4.01 3.20 2.26 1.75 3.09 K=7
52
+ verify 10: accepted 4/6 (K=7) + correction -> +5 tokens (185.7 ms) cpos=32
53
+ [margins] 7.40 0.17 3.37 0.70 4.20 3.20 K=2
54
+ verify 11: accepted 1/1 (K=2) + correction -> +2 tokens (100.1 ms) cpos=34
55
+ [margins] 2.14 4.04 4.80 1.53 1.47 3.93 K=5
56
+ verify 12: accepted 4/4 (K=5) + correction -> +5 tokens (144.6 ms) cpos=39
57
+ [margins] 6.32 0.94 4.36 3.99 6.17 3.20 K=2
58
+ verify 13: accepted 1/1 (K=2) + correction -> +2 tokens (107.2 ms) cpos=41
59
+ [margins] 3.19 4.55 5.26 0.27 1.05 2.19 K=4
60
+ verify 14: accepted 3/3 (K=4) + correction -> +4 tokens (133.4 ms) cpos=45
61
+ [margins] 2.87 1.23 7.06 2.45 5.44 4.76 K=2
62
+ verify 15: accepted 1/1 (K=2) + correction -> +2 tokens (100.4 ms) cpos=47
63
+ [margins] 1.09 4.18 2.70 3.21 3.79 3.45 K=7
64
+ verify 16: accepted 2/6 (K=7) + correction -> +3 tokens (184.7 ms) cpos=50
65
+ [margins] 4.10 4.29 3.32 1.21 3.30 2.90 K=4
66
+ verify 17: accepted 3/3 (K=4) + correction -> +4 tokens (135.9 ms) cpos=54
67
+ [margins] 1.99 2.70 5.70 0.40 3.08 3.12 K=4
68
+ verify 18: accepted 3/3 (K=4) + correction -> +4 tokens (133.1 ms) cpos=58
69
+ [margins] 0.99 4.19 6.23 4.59 0.84 3.46 K=5
70
+ verify 19: accepted 4/4 (K=5) + correction -> +5 tokens (146.4 ms) cpos=63
71
+ [margins] 0.30 5.68 0.64 0.46 4.96 3.56 K=3
72
+ verify 20: accepted 2/2 (K=3) + correction -> +3 tokens (120.9 ms) cpos=66
73
+ [margins] 4.24 7.21 7.27 3.04 1.79 4.27 K=7
74
+ verify 21: accepted 4/6 (K=7) + correction -> +5 tokens (190.1 ms) cpos=71
75
+ [margins] 6.51 1.59 3.97 2.93 4.53 4.91 K=7
76
+ verify 22: accepted 6/6 (K=7) + correction -> +7 tokens (191.6 ms) cpos=78
77
+ [margins] 6.02 1.58 4.42 6.01 4.54 3.51 K=7
78
+ verify 23: accepted 1/6 (K=7) + correction -> +2 tokens (193.3 ms) cpos=80
79
+ [margins] 5.03 6.92 4.40 2.60 2.96 2.51 K=7
80
+ verify 24: accepted 4/6 (K=7) + correction -> +5 tokens (185.0 ms) cpos=85
81
+ [margins] 6.27 0.76 0.37 7.64 4.59 4.76 K=2
82
+ verify 25: accepted 1/1 (K=2) + correction -> +2 tokens (104.2 ms) cpos=87
83
+ [margins] 5.10 6.66 6.83 2.34 1.16 1.34 K=5
84
+ verify 26: accepted 4/4 (K=5) + correction -> +5 tokens (157.1 ms) cpos=92
85
+ [margins] 7.96 0.60 3.49 2.87 3.92 2.34 K=2
86
+ verify 27: accepted 1/1 (K=2) + correction -> +2 tokens (105.2 ms) cpos=94
87
+ [margins] 5.11 5.92 2.43 2.39 0.48 0.03 K=5
88
+ verify 28: accepted 4/4 (K=5) + correction -> +5 tokens (150.9 ms) cpos=99
89
+ [margins] 7.81 1.95 2.97 0.52 6.71 2.92 K=4
90
+ verify 29: accepted 1/3 (K=4) + correction -> +2 tokens (139.9 ms) cpos=101
91
+ [margins] 5.02 4.89 2.33 1.37 4.96 4.17 K=4
92
+ verify 30: accepted 3/3 (K=4) + correction -> +4 tokens (138.6 ms) cpos=105
93
+ [margins] 1.84 6.77 0.62 5.21 4.48 5.20 K=3
94
+ verify 31: accepted 2/2 (K=3) + correction -> +3 tokens (125.4 ms) cpos=108
95
+ [margins] 6.96 2.98 6.09 1.81 1.66 1.05 K=6
96
+ verify 32: accepted 4/5 (K=6) + correction -> +5 tokens (169.0 ms) cpos=113
97
+ [margins] 8.51 0.07 2.95 2.04 5.61 2.47 K=2
98
+ verify 33: accepted 1/1 (K=2) + correction -> +2 tokens (106.5 ms) cpos=115
99
+ [margins] 2.57 1.80 7.69 2.44 1.01 0.94 K=5
100
+ verify 34: accepted 4/4 (K=5) + correction -> +5 tokens (159.1 ms) cpos=120
101
+ [margins] 7.97 0.44 4.16 5.55 5.66 2.86 K=2
102
+ verify 35: accepted 1/1 (K=2) + correction -> +2 tokens (106.7 ms) cpos=122
103
+ [margins] 6.51 3.83 6.41 1.94 0.34 4.99 K=5
104
+ verify 36: accepted 4/4 (K=5) + correction -> +5 tokens (153.1 ms) cpos=127
105
+ [margins] 9.73 0.29 3.74 4.12 5.46 3.86 K=2
106
+ verify 37: accepted 1/1 (K=2) + correction -> +2 tokens (154.4 ms) cpos=129
107
+ [margins] 6.88 4.95 7.68 0.39 1.60 5.89 K=4
108
+ verify 38: accepted 3/3 (K=4) + correction -> +4 tokens (140.2 ms) cpos=133
109
+ [margins] 0.33 4.66 2.69 0.72 0.97 1.48 K=4
110
+ verify 39: accepted 0/3 (K=4) + correction -> +1 tokens (140.0 ms) cpos=134
111
+ [margins] 7.68 0.21 4.24 0.04 8.17 3.66 K=2
112
+ verify 40: accepted 1/1 (K=2) + correction -> +2 tokens (115.8 ms) cpos=136
113
+ [margins] 6.91 6.96 5.99 1.24 1.61 0.03 K=4
114
+ verify 41: accepted 3/3 (K=4) + correction -> +4 tokens (141.2 ms) cpos=140
115
+ [margins] 0.20 3.93 1.26 1.45 0.11 2.57 K=3
116
+ verify 42: accepted 0/2 (K=3) + correction -> +1 tokens (120.5 ms) cpos=141
117
+ [margins] 5.72 7.11 4.21 0.78 1.51 2.88 K=4
118
+ verify 43: accepted 3/3 (K=4) + correction -> +4 tokens (139.3 ms) cpos=145
119
+ [margins] 8.20 6.17 5.97 1.78 0.59 4.59 K=5
120
+ verify 44: accepted 4/4 (K=5) + correction -> +5 tokens (151.8 ms) cpos=150
121
+ [margins] 7.76 0.23 4.14 5.75 6.25 2.19 K=2
122
+ verify 45: accepted 1/1 (K=2) + correction -> +2 tokens (114.3 ms) cpos=152
123
+ [margins] 14.41 3.38 5.26 7.34 8.30 1.86 K=7
124
+ verify 46: accepted 1/6 (K=7) + correction -> +2 tokens (190.8 ms) cpos=154
125
+ [margins] 3.89 5.05 7.95 5.64 1.72 2.79 K=7
126
+ verify 47: accepted 4/6 (K=7) + correction -> +5 tokens (196.8 ms) cpos=159
127
+ [margins] 7.25 0.45 2.09 5.67 8.14 5.76 K=2
128
+ verify 48: accepted 1/1 (K=2) + correction -> +2 tokens (114.7 ms) cpos=161
129
+ [margins] 11.47 6.14 8.44 7.99 2.51 1.43 K=6
130
+ verify 49: accepted 5/5 (K=6) + correction -> +6 tokens (170.7 ms) cpos=167
131
+ [margins] 7.56 0.68 5.74 3.28 1.67 2.16 K=2
132
+ verify 50: accepted 1/1 (K=2) + correction -> +2 tokens (114.4 ms) cpos=169
133
+ [margins] 2.53 2.69 5.46 0.32 0.70 1.71 K=4
134
+ verify 51: accepted 3/3 (K=4) + correction -> +4 tokens (140.2 ms) cpos=173
135
+ [margins] 1.60 4.21 1.11 4.50 5.50 5.99 K=3
136
+ verify 52: accepted 2/2 (K=3) + correction -> +3 tokens (127.3 ms) cpos=176
137
+ [margins] 2.70 2.61 5.59 1.68 0.87 2.08 K=5
138
+ verify 53: accepted 4/4 (K=5) + correction -> +5 tokens (152.4 ms) cpos=181
139
+ [margins] 7.81 1.18 6.14 5.37 5.02 4.41 K=2
140
+ verify 54: accepted 1/1 (K=2) + correction -> +2 tokens (107.7 ms) cpos=183
141
+ [margins] 3.52 5.85 7.09 2.13 0.34 1.67 K=5
142
+ verify 55: accepted 4/4 (K=5) + correction -> +5 tokens (160.7 ms) cpos=188
143
+ [margins] 0.83 2.17 0.77 0.06 8.76 5.71 K=3
144
+ verify 56: accepted 1/2 (K=3) + correction -> +2 tokens (119.8 ms) cpos=190
145
+ [margins] 8.07 1.68 4.68 4.49 3.48 1.84 K=7
146
+ verify 57: accepted 1/6 (K=7) + correction -> +2 tokens (198.8 ms) cpos=192
147
+ [margins] 4.62 3.17 6.55 5.70 7.37 1.10 K=6
148
+ verify 58: accepted 5/5 (K=6) + correction -> +6 tokens (170.2 ms) cpos=198
149
+ [margins] 1.47 5.26 1.08 2.64 4.21 4.18 K=3
150
+ verify 59: accepted 2/2 (K=3) + correction -> +3 tokens (128.4 ms) cpos=201
151
+ [margins] 6.00 6.88 5.76 4.47 0.53 0.68 K=5
152
+ verify 60: accepted 4/4 (K=5) + correction -> +5 tokens (153.1 ms) cpos=206
153
+
154
+ [spec] generated 201 tokens over 60 verifies: mean tokens/verify = 3.35 (block=6, max 7)
155
+ [spec] tokens: 11111 16 455 6102 294 16603 344 29168 16 455 6102 294 14251 344 16235 16 455 6102 294 29585 344 76405 16 455 6102 294 25062 344 40977 16 455 6102 294 10322 344 17575 16 455 6102 294
156
+ [spec] LOSSLESS GATE: first 8 tokens match base AR -> PASS
157
+ [spec] SPEC-DECODE: 42.0 ms/tok = 23.82 tok/s (vs base M=1 70.7 ms/tok = 14.14 tok/s -> 1.68x)
158
+ [decode] PREFILL: 34 positions in 624.2 ms = 54.5 tok/s (18.358 ms/tok)
159
+ [spec] decoding 200 tokens (block=6, draft passes=1, adaptK=1.50, prompt=1 s=35)...
160
+ [margins] 0.84 0.66 0.14 0.11 0.37 0.98 K=2
161
+ verify 1: accepted 0/1 (K=2) + correction -> +1 tokens (109.9 ms) cpos=35
162
+ [margins] 0.39 1.76 1.20 2.13 4.90 0.63 K=3
163
+ verify 2: accepted 0/2 (K=3) + correction -> +1 tokens (118.6 ms) cpos=36
164
+ [margins] 1.22 1.77 1.08 3.93 1.37 1.44 K=3
165
+ verify 3: accepted 0/2 (K=3) + correction -> +1 tokens (110.0 ms) cpos=37
166
+ [margins] 3.11 3.04 0.49 1.24 1.04 1.21 K=3
167
+ verify 4: accepted 2/2 (K=3) + correction -> +3 tokens (117.0 ms) cpos=40
168
+ [margins] 0.57 0.64 0.02 1.55 1.86 0.60 K=2
169
+ verify 5: accepted 0/1 (K=2) + correction -> +1 tokens (100.2 ms) cpos=41
170
+ [margins] 1.33 4.56 1.01 0.01 0.13 2.59 K=3
171
+ verify 6: accepted 2/2 (K=3) + correction -> +3 tokens (118.6 ms) cpos=44
172
+ [margins] 0.15 0.64 1.71 6.89 1.02 0.80 K=2
173
+ verify 7: accepted 0/1 (K=2) + correction -> +1 tokens (101.3 ms) cpos=45
174
+ [margins] 7.45 12.64 3.81 3.00 0.92 0.22 K=5
175
+ verify 8: accepted 4/4 (K=5) + correction -> +5 tokens (143.4 ms) cpos=50
176
+ [margins] 3.62 2.25 1.29 1.94 1.34 4.57 K=3
177
+ verify 9: accepted 1/2 (K=3) + correction -> +2 tokens (119.1 ms) cpos=52
178
+ [margins] 3.76 4.86 7.49 2.11 1.05 1.94 K=5
179
+ verify 10: accepted 3/4 (K=5) + correction -> +4 tokens (147.4 ms) cpos=56
180
+ [margins] 2.42 0.55 1.25 0.15 0.50 0.07 K=2
181
+ verify 11: accepted 1/1 (K=2) + correction -> +2 tokens (105.4 ms) cpos=58
182
+ [margins] 3.00 3.39 2.49 0.65 0.07 2.72 K=4
183
+ verify 12: accepted 0/3 (K=4) + correction -> +1 tokens (123.0 ms) cpos=59
184
+ [margins] 0.51 0.78 2.66 0.55 5.11 0.46 K=2
185
+ verify 13: accepted 0/1 (K=2) + correction -> +1 tokens (108.9 ms) cpos=60
186
+ [margins] 1.44 2.08 5.42 0.19 0.84 0.80 K=4
187
+ verify 14: accepted 0/3 (K=4) + correction -> +1 tokens (133.0 ms) cpos=61
188
+ [margins] 1.32 4.66 0.32 2.14 0.54 0.70 K=3
189
+ verify 15: accepted 1/2 (K=3) + correction -> +2 tokens (121.1 ms) cpos=63
190
+ [margins] 1.03 5.03 0.21 1.59 0.15 0.82 K=3
191
+ verify 16: accepted 0/2 (K=3) + correction -> +1 tokens (122.2 ms) cpos=64
192
+ [margins] 5.09 0.37 1.78 5.47 0.20 0.18 K=2
193
+ verify 17: accepted 1/1 (K=2) + correction -> +2 tokens (101.0 ms) cpos=66
194
+ [margins] 1.74 2.54 2.89 4.96 0.35 0.42 K=5
195
+ verify 18: accepted 4/4 (K=5) + correction -> +5 tokens (145.8 ms) cpos=71
196
+ [margins] 2.93 1.68 1.52 5.33 0.18 1.93 K=5
197
+ verify 19: accepted 4/4 (K=5) + correction -> +5 tokens (154.0 ms) cpos=76
198
+ [margins] 5.17 4.63 1.66 6.91 2.37 2.63 K=7
199
+ verify 20: accepted 6/6 (K=7) + correction -> +7 tokens (177.7 ms) cpos=83
200
+ [margins] 0.15 2.22 0.13 0.81 1.12 1.60 K=3
201
+ verify 21: accepted 0/2 (K=3) + correction -> +1 tokens (127.3 ms) cpos=84
202
+ [margins] 7.44 3.67 5.60 3.94 4.18 3.43 K=7
203
+ verify 22: accepted 6/6 (K=7) + correction -> +7 tokens (182.7 ms) cpos=91
204
+ [margins] 6.36 3.04 3.79 7.68 1.87 3.12 K=7
205
+ verify 23: accepted 6/6 (K=7) + correction -> +7 tokens (188.5 ms) cpos=98
206
+ [margins] 3.55 7.50 3.25 7.03 6.06 4.52 K=7
207
+ verify 24: accepted 6/6 (K=7) + correction -> +7 tokens (190.3 ms) cpos=105
208
+ [margins] 7.67 7.37 8.63 1.03 6.59 0.41 K=4
209
+ verify 25: accepted 3/3 (K=4) + correction -> +4 tokens (138.1 ms) cpos=109
210
+ [margins] 8.80 7.66 5.36 8.76 6.91 5.95 K=7
211
+ verify 26: accepted 6/6 (K=7) + correction -> +7 tokens (192.0 ms) cpos=116
212
+ [margins] 9.41 9.52 4.84 9.43 4.76 6.77 K=7
213
+ verify 27: accepted 6/6 (K=7) + correction -> +7 tokens (187.2 ms) cpos=123
214
+ [margins] 4.11 7.37 8.26 7.46 4.21 2.98 K=7
215
+ verify 28: accepted 6/6 (K=7) + correction -> +7 tokens (236.3 ms) cpos=130
216
+ [margins] 8.38 8.36 5.40 3.62 8.00 0.34 K=6
217
+ verify 29: accepted 5/5 (K=6) + correction -> +6 tokens (175.2 ms) cpos=136
218
+ [margins] 8.40 6.28 6.82 8.56 5.64 4.74 K=7
219
+ verify 30: accepted 6/6 (K=7) + correction -> +7 tokens (190.0 ms) cpos=143
220
+ [margins] 2.25 7.29 9.76 9.19 6.36 6.86 K=7
221
+ verify 31: accepted 6/6 (K=7) + correction -> +7 tokens (197.0 ms) cpos=150
222
+ [margins] 8.97 8.90 8.90 4.58 7.22 3.65 K=7
223
+ verify 32: accepted 6/6 (K=7) + correction -> +7 tokens (196.5 ms) cpos=157
224
+ [margins] 11.81 2.92 7.54 8.65 6.62 7.16 K=7
225
+ verify 33: accepted 6/6 (K=7) + correction -> +7 tokens (195.8 ms) cpos=164
226
+ [margins] 9.72 11.47 8.43 5.40 8.02 8.43 K=7
227
+ verify 34: accepted 6/6 (K=7) + correction -> +7 tokens (191.1 ms) cpos=171
228
+ [margins] 9.73 3.52 5.06 2.51 3.64 2.29 K=7
229
+ verify 35: accepted 6/6 (K=7) + correction -> +7 tokens (197.0 ms) cpos=178
230
+ [margins] 5.36 9.38 5.78 9.49 7.24 7.88 K=7
231
+ verify 36: accepted 6/6 (K=7) + correction -> +7 tokens (193.7 ms) cpos=185
232
+ [margins] 11.44 12.07 12.62 10.49 7.56 10.66 K=7
233
+ verify 37: accepted 6/6 (K=7) + correction -> +7 tokens (196.2 ms) cpos=192
234
+ [margins] 12.73 6.87 9.54 10.72 8.30 2.27 K=7
235
+ verify 38: accepted 6/6 (K=7) + correction -> +7 tokens (189.5 ms) cpos=199
236
+ [margins] 10.24 9.35 8.66 6.37 8.18 5.60 K=7
237
+ verify 39: accepted 6/6 (K=7) + correction -> +7 tokens (197.9 ms) cpos=206
238
+ [margins] 11.14 7.79 4.68 1.22 1.26 0.11 K=4
239
+ verify 40: accepted 3/3 (K=4) + correction -> +4 tokens (142.8 ms) cpos=210
240
+ [margins] 9.25 11.49 10.25 7.81 6.31 10.28 K=7
241
+ verify 41: accepted 6/6 (K=7) + correction -> +7 tokens (198.4 ms) cpos=217
242
+ [margins] 14.56 5.26 9.67 11.01 10.25 5.52 K=7
243
+ verify 42: accepted 6/6 (K=7) + correction -> +7 tokens (197.9 ms) cpos=224
244
+ [margins] 11.60 8.76 9.08 4.68 8.63 1.84 K=7
245
+ verify 43: accepted 6/6 (K=7) + correction -> +7 tokens (191.1 ms) cpos=231
246
+ [margins] 9.32 8.61 9.32 9.36 4.40 2.72 K=7
247
+ verify 44: accepted 6/6 (K=7) + correction -> +7 tokens (200.7 ms) cpos=238
248
+
249
+ [spec] generated 204 tokens over 44 verifies: mean tokens/verify = 4.64 (block=6, max 7)
250
+ [spec] tokens: 4147 554 1926 270 4105 339 43 1114 1926 270 1178 65 50219 4105 418 270 9517 28 4593 1281 106239 3305 1 1 14 223 19 16 23 14 223 20 16 23 14 223 21 16 23 14
251
+ [spec] SPEC-DECODE: 33.8 ms/tok = 29.59 tok/s (vs base M=1 70.7 ms/tok = 14.14 tok/s -> 2.09x)
252
+ [decode] PREFILL: 41 positions in 723.0 ms = 56.7 tok/s (17.634 ms/tok)
253
+ [spec] decoding 200 tokens (block=6, draft passes=1, adaptK=1.50, prompt=2 s=42)...
254
+ [margins] 5.86 0.27 1.45 2.34 1.33 0.19 K=2
255
+ verify 1: accepted 1/1 (K=2) + correction -> +2 tokens (104.0 ms) cpos=43
256
+ [margins] 3.54 1.15 0.04 0.20 0.88 0.50 K=2
257
+ verify 2: accepted 1/1 (K=2) + correction -> +2 tokens (101.0 ms) cpos=45
258
+ [margins] 1.93 3.41 1.82 0.27 0.27 0.36 K=4
259
+ verify 3: accepted 2/3 (K=4) + correction -> +3 tokens (121.1 ms) cpos=48
260
+ [margins] 3.33 5.44 0.51 1.24 3.75 0.15 K=3
261
+ verify 4: accepted 0/2 (K=3) + correction -> +1 tokens (113.5 ms) cpos=49
262
+ [margins] 2.72 6.79 1.55 0.10 1.34 4.70 K=4
263
+ verify 5: accepted 0/3 (K=4) + correction -> +1 tokens (129.8 ms) cpos=50
264
+ [margins] 3.91 3.78 1.87 0.57 0.97 3.77 K=4
265
+ verify 6: accepted 0/3 (K=4) + correction -> +1 tokens (121.2 ms) cpos=51
266
+ [margins] 8.58 3.42 0.15 1.66 3.72 1.43 K=3
267
+ verify 7: accepted 1/2 (K=3) + correction -> +2 tokens (116.4 ms) cpos=53
268
+ [margins] 2.19 5.83 2.32 0.31 2.29 4.16 K=4
269
+ verify 8: accepted 0/3 (K=4) + correction -> +1 tokens (129.6 ms) cpos=54
270
+ [margins] 3.86 9.35 3.14 2.32 1.82 2.41 K=7
271
+ verify 9: accepted 2/6 (K=7) + correction -> +3 tokens (180.0 ms) cpos=57
272
+ [margins] 5.13 9.34 1.82 0.07 5.27 4.47 K=4
273
+ verify 10: accepted 0/3 (K=4) + correction -> +1 tokens (130.6 ms) cpos=58
274
+ [margins] 5.62 6.88 0.35 2.65 3.86 3.88 K=3
275
+ verify 11: accepted 2/2 (K=3) + correction -> +3 tokens (120.9 ms) cpos=61
276
+ [margins] 0.13 7.69 12.25 0.21 1.62 8.13 K=4
277
+ verify 12: accepted 3/3 (K=4) + correction -> +4 tokens (135.6 ms) cpos=65
278
+ [margins] 2.11 7.55 9.85 0.19 1.89 7.22 K=4
279
+ verify 13: accepted 3/3 (K=4) + correction -> +4 tokens (135.6 ms) cpos=69
280
+ [margins] 5.53 7.17 8.90 0.45 6.05 6.80 K=4
281
+ verify 14: accepted 2/3 (K=4) + correction -> +3 tokens (135.6 ms) cpos=72
282
+ [margins] 4.94 2.22 5.95 0.42 2.50 8.88 K=4
283
+ verify 15: accepted 1/3 (K=4) + correction -> +2 tokens (134.3 ms) cpos=74
284
+ [margins] 4.94 2.80 2.86 0.30 7.94 1.51 K=4
285
+ verify 16: accepted 1/3 (K=4) + correction -> +2 tokens (133.0 ms) cpos=76
286
+ [margins] 11.29 4.72 5.42 4.01 0.52 3.42 K=5
287
+ verify 17: accepted 2/4 (K=5) + correction -> +3 tokens (146.8 ms) cpos=79
288
+ [margins] 16.56 8.77 4.66 6.03 2.68 10.53 K=7
289
+ verify 18: accepted 6/6 (K=7) + correction -> +7 tokens (187.3 ms) cpos=86
290
+ [margins] 6.97 8.32 7.29 6.20 0.70 5.10 K=5
291
+ verify 19: accepted 4/4 (K=5) + correction -> +5 tokens (145.8 ms) cpos=91
292
+ [margins] 14.22 12.41 3.42 6.43 10.65 6.02 K=7
293
+ verify 20: accepted 6/6 (K=7) + correction -> +7 tokens (187.2 ms) cpos=98
294
+ [margins] 10.80 7.90 7.53 4.19 1.33 4.07 K=5
295
+ verify 21: accepted 4/4 (K=5) + correction -> +5 tokens (148.6 ms) cpos=103
296
+ [margins] 14.09 13.15 9.45 0.93 2.74 6.35 K=4
297
+ verify 22: accepted 3/3 (K=4) + correction -> +4 tokens (140.4 ms) cpos=107
298
+ [margins] 12.89 13.31 7.41 5.08 2.71 0.75 K=6
299
+ verify 23: accepted 4/5 (K=6) + correction -> +5 tokens (172.1 ms) cpos=112
300
+ [margins] 15.34 9.11 6.90 5.62 0.50 4.60 K=5
301
+ verify 24: accepted 2/4 (K=5) + correction -> +3 tokens (150.4 ms) cpos=115
302
+ [margins] 12.38 9.55 10.50 2.01 2.66 14.22 K=7
303
+ verify 25: accepted 3/6 (K=7) + correction -> +4 tokens (193.3 ms) cpos=119
304
+ [margins] 5.95 6.46 1.52 5.40 0.43 1.16 K=5
305
+ verify 26: accepted 4/4 (K=5) + correction -> +5 tokens (157.9 ms) cpos=124
306
+ [margins] 10.30 14.03 9.23 5.78 1.34 0.52 K=5
307
+ verify 27: accepted 4/4 (K=5) + correction -> +5 tokens (190.8 ms) cpos=129
308
+ [margins] 9.31 9.18 9.34 3.33 3.69 5.17 K=7
309
+ verify 28: accepted 6/6 (K=7) + correction -> +7 tokens (196.0 ms) cpos=136
310
+ [margins] 15.12 11.06 17.82 7.96 10.42 3.05 K=7
311
+ verify 29: accepted 6/6 (K=7) + correction -> +7 tokens (187.1 ms) cpos=143
312
+ [margins] 9.97 10.70 7.85 4.74 1.33 0.96 K=5
313
+ verify 30: accepted 4/4 (K=5) + correction -> +5 tokens (158.2 ms) cpos=148
314
+ [margins] 17.74 13.76 11.21 5.31 4.86 1.10 K=6
315
+ verify 31: accepted 5/5 (K=6) + correction -> +6 tokens (169.0 ms) cpos=154
316
+ [margins] 7.65 12.88 8.68 6.25 1.06 10.19 K=5
317
+ verify 32: accepted 3/4 (K=5) + correction -> +4 tokens (152.9 ms) cpos=158
318
+ [margins] 11.79 7.09 9.37 0.07 3.99 1.34 K=4
319
+ verify 33: accepted 3/3 (K=4) + correction -> +4 tokens (137.3 ms) cpos=162
320
+ [margins] 16.08 0.42 3.89 4.59 0.73 2.14 K=2
321
+ verify 34: accepted 1/1 (K=2) + correction -> +2 tokens (112.6 ms) cpos=164
322
+ [margins] 11.26 7.71 0.19 2.74 3.59 0.32 K=3
323
+ verify 35: accepted 2/2 (K=3) + correction -> +3 tokens (122.0 ms) cpos=167
324
+ [margins] 9.38 14.27 13.28 7.08 0.17 1.54 K=5
325
+ verify 36: accepted 4/4 (K=5) + correction -> +5 tokens (159.1 ms) cpos=172
326
+ [margins] 14.02 9.27 7.04 5.85 0.02 1.40 K=5
327
+ verify 37: accepted 4/4 (K=5) + correction -> +5 tokens (152.9 ms) cpos=177
328
+ [margins] 12.20 15.22 12.81 6.46 6.09 5.86 K=7
329
+ verify 38: accepted 6/6 (K=7) + correction -> +7 tokens (194.3 ms) cpos=184
330
+ [margins] 18.98 10.66 4.51 12.97 8.27 5.28 K=7
331
+ verify 39: accepted 6/6 (K=7) + correction -> +7 tokens (190.3 ms) cpos=191
332
+ [margins] 7.33 13.72 2.69 0.39 3.99 1.04 K=4
333
+ verify 40: accepted 2/3 (K=4) + correction -> +3 tokens (141.9 ms) cpos=194
334
+ [margins] 19.22 9.43 4.27 14.79 6.76 10.76 K=7
335
+ verify 41: accepted 6/6 (K=7) + correction -> +7 tokens (198.1 ms) cpos=201
336
+ [margins] 7.94 11.32 7.32 6.38 3.37 3.02 K=7
337
+ verify 42: accepted 5/6 (K=7) + correction -> +6 tokens (196.5 ms) cpos=207
338
+ [margins] 10.17 7.71 8.32 5.50 4.18 5.08 K=7
339
+ verify 43: accepted 5/6 (K=7) + correction -> +6 tokens (192.5 ms) cpos=213
340
+ [margins] 17.31 11.96 7.16 14.83 7.37 7.71 K=7
341
+ verify 44: accepted 6/6 (K=7) + correction -> +7 tokens (195.8 ms) cpos=220
342
+ [margins] 8.01 18.14 4.21 14.21 4.63 1.61 K=7
343
+ verify 45: accepted 6/6 (K=7) + correction -> +7 tokens (190.0 ms) cpos=227
344
+ [margins] 11.98 18.03 10.63 8.35 11.94 3.92 K=7
345
+ verify 46: accepted 5/6 (K=7) + correction -> +6 tokens (199.5 ms) cpos=233
346
+ [margins] 17.22 12.74 6.84 11.72 11.51 5.56 K=7
347
+ verify 47: accepted 6/6 (K=7) + correction -> +7 tokens (200.5 ms) cpos=240
348
+ [margins] 8.61 19.15 0.72 1.11 1.59 6.68 K=3
349
+ verify 48: accepted 2/2 (K=3) + correction -> +3 tokens (125.2 ms) cpos=243
350
+
351
+ [spec] generated 202 tokens over 48 verifies: mean tokens/verify = 4.21 (block=6, max 7)
352
+ [spec] tokens: 1 1 1 1 1 1 50 1697 1 16 1 50 1697 16 1 50 1697 16 1 50 1697 16 1 50 1697 16 1 50 1697 16 20692 50 1697 16 791 50 1697 4394 1015 50
353
+ [spec] SPEC-DECODE: 36.7 ms/tok = 27.28 tok/s (vs base M=1 70.7 ms/tok = 14.14 tok/s -> 1.93x)
354
+ [decode] PREFILL: 26 positions in 477.8 ms = 54.4 tok/s (18.378 ms/tok)
355
+ [spec] decoding 200 tokens (block=6, draft passes=1, adaptK=1.50, prompt=3 s=27)...
356
+ [margins] 1.44 0.66 2.32 0.53 0.54 6.68 K=2
357
+ verify 1: accepted 1/1 (K=2) + correction -> +2 tokens (109.5 ms) cpos=28
358
+ [margins] 4.35 2.95 0.07 0.95 0.98 0.04 K=3
359
+ verify 2: accepted 1/2 (K=3) + correction -> +2 tokens (113.5 ms) cpos=30
360
+ [margins] 2.62 0.14 0.61 0.00 1.51 1.70 K=2
361
+ verify 3: accepted 1/1 (K=2) + correction -> +2 tokens (108.5 ms) cpos=32
362
+ [margins] 2.21 0.24 0.08 0.29 0.26 3.77 K=2
363
+ verify 4: accepted 1/1 (K=2) + correction -> +2 tokens (102.4 ms) cpos=34
364
+ [margins] 2.89 0.20 1.04 2.09 0.35 1.09 K=2
365
+ verify 5: accepted 1/1 (K=2) + correction -> +2 tokens (108.1 ms) cpos=36
366
+ [margins] 2.98 0.84 3.00 1.60 0.94 0.09 K=2
367
+ verify 6: accepted 1/1 (K=2) + correction -> +2 tokens (102.6 ms) cpos=38
368
+ [margins] 3.11 0.66 3.68 1.27 0.86 0.96 K=2
369
+ verify 7: accepted 1/1 (K=2) + correction -> +2 tokens (108.6 ms) cpos=40
370
+ [margins] 2.95 1.39 3.85 1.61 1.02 1.51 K=2
371
+ verify 8: accepted 1/1 (K=2) + correction -> +2 tokens (102.8 ms) cpos=42
372
+ [margins] 2.91 0.57 4.52 2.15 0.14 1.91 K=2
373
+ verify 9: accepted 1/1 (K=2) + correction -> +2 tokens (108.8 ms) cpos=44
374
+ [margins] 2.91 0.25 9.39 2.47 0.30 7.58 K=2
375
+ verify 10: accepted 1/1 (K=2) + correction -> +2 tokens (102.9 ms) cpos=46
376
+ [margins] 3.04 0.54 7.90 2.39 0.09 7.70 K=2
377
+ verify 11: accepted 1/1 (K=2) + correction -> +2 tokens (109.2 ms) cpos=48
378
+ [margins] 2.80 0.01 11.19 4.00 1.04 8.09 K=2
379
+ verify 12: accepted 1/1 (K=2) + correction -> +2 tokens (102.6 ms) cpos=50
380
+ [margins] 2.04 0.26 9.38 2.88 0.84 8.28 K=2
381
+ verify 13: accepted 1/1 (K=2) + correction -> +2 tokens (109.2 ms) cpos=52
382
+ [margins] 3.01 0.42 10.48 3.17 0.46 8.52 K=2
383
+ verify 14: accepted 1/1 (K=2) + correction -> +2 tokens (103.4 ms) cpos=54
384
+ [margins] 3.65 0.67 10.46 3.70 0.67 8.60 K=2
385
+ verify 15: accepted 1/1 (K=2) + correction -> +2 tokens (109.4 ms) cpos=56
386
+ [margins] 3.42 0.69 10.38 4.05 1.04 8.68 K=2
387
+ verify 16: accepted 1/1 (K=2) + correction -> +2 tokens (103.0 ms) cpos=58
388
+ [margins] 3.69 0.10 11.43 3.68 1.19 8.57 K=2
389
+ verify 17: accepted 1/1 (K=2) + correction -> +2 tokens (109.7 ms) cpos=60
390
+ [margins] 3.55 0.46 10.42 3.50 0.89 8.86 K=2
391
+ verify 18: accepted 1/1 (K=2) + correction -> +2 tokens (103.7 ms) cpos=62
392
+ [margins] 3.84 0.20 9.44 3.90 0.40 8.99 K=2
393
+ verify 19: accepted 1/1 (K=2) + correction -> +2 tokens (110.4 ms) cpos=64
394
+ [margins] 4.30 0.98 10.47 4.49 0.80 8.63 K=2
395
+ verify 20: accepted 1/1 (K=2) + correction -> +2 tokens (103.2 ms) cpos=66
396
+ [margins] 3.71 0.14 10.40 4.37 0.75 8.60 K=2
397
+ verify 21: accepted 1/1 (K=2) + correction -> +2 tokens (110.3 ms) cpos=68
398
+ [margins] 3.89 0.21 11.56 4.24 1.10 8.75 K=2
399
+ verify 22: accepted 1/1 (K=2) + correction -> +2 tokens (103.8 ms) cpos=70
400
+ [margins] 3.98 0.43 11.35 4.20 0.92 8.74 K=2
401
+ verify 23: accepted 1/1 (K=2) + correction -> +2 tokens (110.7 ms) cpos=72
402
+ [margins] 3.47 1.11 10.09 4.37 0.53 8.67 K=2
403
+ verify 24: accepted 1/1 (K=2) + correction -> +2 tokens (103.9 ms) cpos=74
404
+ [margins] 3.68 0.28 10.43 4.17 0.67 8.76 K=2
405
+ verify 25: accepted 0/1 (K=2) + correction -> +1 tokens (110.3 ms) cpos=75
406
+ [margins] 2.85 1.26 0.03 8.41 1.97 0.22 K=2
407
+ verify 26: accepted 0/1 (K=2) + correction -> +1 tokens (109.0 ms) cpos=76
408
+ [margins] 1.88 8.76 3.16 1.53 2.33 0.15 K=6
409
+ verify 27: accepted 0/5 (K=6) + correction -> +1 tokens (161.1 ms) cpos=77
410
+ [margins] 5.58 3.70 0.82 0.75 0.65 0.03 K=3
411
+ verify 28: accepted 2/2 (K=3) + correction -> +3 tokens (120.5 ms) cpos=80
412
+ [margins] 0.76 2.67 1.59 3.32 0.37 1.48 K=5
413
+ verify 29: accepted 0/4 (K=5) + correction -> +1 tokens (143.1 ms) cpos=81
414
+ [margins] 2.51 1.46 1.13 1.37 1.00 1.07 K=2
415
+ verify 30: accepted 1/1 (K=2) + correction -> +2 tokens (101.7 ms) cpos=83
416
+ [margins] 0.21 3.38 1.16 1.75 0.43 1.41 K=3
417
+ verify 31: accepted 0/2 (K=3) + correction -> +1 tokens (120.3 ms) cpos=84
418
+ [margins] 5.47 4.02 0.32 1.34 0.02 2.05 K=3
419
+ verify 32: accepted 2/2 (K=3) + correction -> +3 tokens (112.3 ms) cpos=87
420
+ [margins] 3.06 2.76 0.95 4.21 0.58 2.53 K=3
421
+ verify 33: accepted 1/2 (K=3) + correction -> +2 tokens (122.5 ms) cpos=89
422
+ [margins] 0.24 3.13 0.25 0.36 0.43 0.69 K=3
423
+ verify 34: accepted 2/2 (K=3) + correction -> +3 tokens (121.2 ms) cpos=92
424
+ [margins] 2.83 1.42 0.85 4.47 2.58 0.03 K=2
425
+ verify 35: accepted 1/1 (K=2) + correction -> +2 tokens (103.7 ms) cpos=94
426
+ [margins] 1.02 9.38 0.92 0.76 2.58 1.42 K=3
427
+ verify 36: accepted 2/2 (K=3) + correction -> +3 tokens (120.2 ms) cpos=97
428
+ [margins] 3.08 2.47 0.23 3.70 0.09 1.01 K=3
429
+ verify 37: accepted 2/2 (K=3) + correction -> +3 tokens (125.3 ms) cpos=100
430
+ [margins] 3.35 1.07 0.80 0.11 1.03 1.14 K=2
431
+ verify 38: accepted 1/1 (K=2) + correction -> +2 tokens (102.2 ms) cpos=102
432
+ [margins] 1.54 0.09 3.53 0.68 7.16 0.56 K=2
433
+ verify 39: accepted 0/1 (K=2) + correction -> +1 tokens (109.0 ms) cpos=103
434
+ [margins] 7.44 5.82 1.26 4.42 5.72 1.93 K=3
435
+ verify 40: accepted 2/2 (K=3) + correction -> +3 tokens (120.2 ms) cpos=106
436
+ [margins] 7.41 4.74 1.76 5.16 0.18 2.62 K=5
437
+ verify 41: accepted 4/4 (K=5) + correction -> +5 tokens (142.4 ms) cpos=111
438
+ [margins] 0.70 6.93 1.59 1.68 0.83 0.21 K=5
439
+ verify 42: accepted 3/4 (K=5) + correction -> +4 tokens (155.1 ms) cpos=115
440
+ [margins] 10.66 0.46 2.65 1.29 1.88 2.55 K=2
441
+ verify 43: accepted 1/1 (K=2) + correction -> +2 tokens (111.9 ms) cpos=117
442
+ [margins] 1.10 6.54 3.96 4.83 9.75 3.99 K=7
443
+ verify 44: accepted 6/6 (K=7) + correction -> +7 tokens (196.7 ms) cpos=124
444
+ [margins] 11.62 3.31 13.72 8.40 4.21 2.33 K=7
445
+ verify 45: accepted 5/6 (K=7) + correction -> +6 tokens (226.2 ms) cpos=130
446
+ [margins] 10.69 9.52 11.43 0.83 0.54 8.78 K=4
447
+ verify 46: accepted 3/3 (K=4) + correction -> +4 tokens (139.4 ms) cpos=134
448
+ [margins] 0.44 2.83 5.12 2.90 2.25 1.02 K=6
449
+ verify 47: accepted 0/5 (K=6) + correction -> +1 tokens (173.7 ms) cpos=135
450
+ [margins] 0.87 0.40 0.58 7.44 0.98 0.99 K=2
451
+ verify 48: accepted 0/1 (K=2) + correction -> +1 tokens (116.6 ms) cpos=136
452
+ [margins] 3.83 3.00 9.43 1.93 1.61 6.56 K=7
453
+ verify 49: accepted 0/6 (K=7) + correction -> +1 tokens (179.9 ms) cpos=137
454
+ [margins] 4.19 2.82 5.47 0.29 0.86 0.16 K=4
455
+ verify 50: accepted 0/3 (K=4) + correction -> +1 tokens (139.8 ms) cpos=138
456
+ [margins] 1.25 0.11 2.30 0.33 1.97 1.22 K=2
457
+ verify 51: accepted 1/1 (K=2) + correction -> +2 tokens (111.9 ms) cpos=140
458
+ [margins] 3.73 3.41 0.91 0.17 0.18 0.31 K=3
459
+ verify 52: accepted 0/2 (K=3) + correction -> +1 tokens (117.9 ms) cpos=141
460
+ [margins] 2.57 0.78 0.21 1.31 1.16 0.82 K=2
461
+ verify 53: accepted 1/1 (K=2) + correction -> +2 tokens (104.6 ms) cpos=143
462
+ [margins] 1.01 0.57 0.52 1.87 0.18 0.68 K=2
463
+ verify 54: accepted 0/1 (K=2) + correction -> +1 tokens (113.3 ms) cpos=144
464
+ [margins] 0.29 1.11 0.62 2.21 0.35 0.00 K=2
465
+ verify 55: accepted 1/1 (K=2) + correction -> +2 tokens (109.3 ms) cpos=146
466
+ [margins] 7.64 3.69 0.37 0.75 1.11 2.92 K=3
467
+ verify 56: accepted 2/2 (K=3) + correction -> +3 tokens (123.4 ms) cpos=149
468
+ [margins] 2.04 3.44 3.88 1.15 2.53 0.30 K=4
469
+ verify 57: accepted 0/3 (K=4) + correction -> +1 tokens (136.5 ms) cpos=150
470
+ [margins] 0.21 0.37 0.96 0.14 2.99 1.17 K=2
471
+ verify 58: accepted 0/1 (K=2) + correction -> +1 tokens (112.0 ms) cpos=151
472
+ [margins] 2.09 2.33 0.16 2.78 0.06 0.65 K=3
473
+ verify 59: accepted 0/2 (K=3) + correction -> +1 tokens (123.7 ms) cpos=152
474
+ [margins] 3.94 1.07 0.23 0.48 2.39 0.34 K=2
475
+ verify 60: accepted 1/1 (K=2) + correction -> +2 tokens (104.7 ms) cpos=154
476
+ [margins] 1.49 0.70 2.21 0.22 1.42 2.26 K=2
477
+ verify 61: accepted 1/1 (K=2) + correction -> +2 tokens (112.8 ms) cpos=156
478
+ [margins] 5.10 1.14 0.06 2.26 0.83 2.13 K=2
479
+ verify 62: accepted 1/1 (K=2) + correction -> +2 tokens (105.6 ms) cpos=158
480
+ [margins] 1.71 2.11 1.91 0.11 0.63 2.36 K=4
481
+ verify 63: accepted 3/3 (K=4) + correction -> +4 tokens (139.7 ms) cpos=162
482
+ [margins] 7.63 1.33 0.56 1.58 1.47 1.67 K=2
483
+ verify 64: accepted 1/1 (K=2) + correction -> +2 tokens (112.9 ms) cpos=164
484
+ [margins] 0.19 0.90 0.85 0.33 2.67 1.51 K=2
485
+ verify 65: accepted 0/1 (K=2) + correction -> +1 tokens (106.5 ms) cpos=165
486
+ [margins] 4.11 1.71 2.40 0.59 0.92 1.52 K=4
487
+ verify 66: accepted 2/3 (K=4) + correction -> +3 tokens (136.9 ms) cpos=168
488
+ [margins] 4.45 0.97 5.93 3.60 1.65 1.14 K=2
489
+ verify 67: accepted 1/1 (K=2) + correction -> +2 tokens (108.5 ms) cpos=170
490
+ [margins] 12.69 3.13 1.55 0.80 0.33 2.99 K=4
491
+ verify 68: accepted 2/3 (K=4) + correction -> +3 tokens (137.8 ms) cpos=173
492
+ [margins] 0.10 0.23 0.92 1.56 1.92 0.45 K=2
493
+ verify 69: accepted 0/1 (K=2) + correction -> +1 tokens (107.1 ms) cpos=174
494
+ [margins] 2.02 2.34 0.96 0.37 0.35 1.94 K=3
495
+ verify 70: accepted 0/2 (K=3) + correction -> +1 tokens (125.2 ms) cpos=175
496
+ [margins] 0.96 2.18 0.38 0.89 1.17 0.51 K=3
497
+ verify 71: accepted 1/2 (K=3) + correction -> +2 tokens (127.9 ms) cpos=177
498
+ [margins] 6.92 3.74 4.59 0.95 2.52 2.30 K=4
499
+ verify 72: accepted 2/3 (K=4) + correction -> +3 tokens (136.3 ms) cpos=180
500
+ [margins] 0.87 2.41 2.69 1.49 0.14 4.50 K=4
501
+ verify 73: accepted 0/3 (K=4) + correction -> +1 tokens (136.7 ms) cpos=181
502
+ [margins] 4.96 0.45 2.75 0.47 1.36 3.86 K=2
503
+ verify 74: accepted 1/1 (K=2) + correction -> +2 tokens (105.9 ms) cpos=183
504
+ [margins] 3.01 4.98 4.56 2.19 2.93 0.27 K=6
505
+ verify 75: accepted 2/5 (K=6) + correction -> +3 tokens (169.8 ms) cpos=186
506
+ [margins] 3.06 0.36 0.24 2.75 0.83 0.31 K=2
507
+ verify 76: accepted 1/1 (K=2) + correction -> +2 tokens (114.8 ms) cpos=188
508
+ [margins] 6.97 5.38 1.50 2.83 3.71 0.21 K=3
509
+ verify 77: accepted 2/2 (K=3) + correction -> +3 tokens (121.0 ms) cpos=191
510
+ [margins] 9.22 1.94 3.58 1.35 3.15 1.14 K=4
511
+ verify 78: accepted 2/3 (K=4) + correction -> +3 tokens (139.8 ms) cpos=194
512
+ [margins] 0.58 0.54 0.10 1.37 0.05 3.28 K=2
513
+ verify 79: accepted 0/1 (K=2) + correction -> +1 tokens (114.9 ms) cpos=195
514
+ [margins] 2.20 0.29 3.22 3.82 0.89 4.78 K=2
515
+ verify 80: accepted 1/1 (K=2) + correction -> +2 tokens (112.8 ms) cpos=197
516
+ [margins] 1.26 2.19 5.29 0.45 3.97 1.13 K=4
517
+ verify 81: accepted 0/3 (K=4) + correction -> +1 tokens (137.4 ms) cpos=198
518
+ [margins] 2.33 0.37 1.93 0.20 0.60 2.06 K=2
519
+ verify 82: accepted 0/1 (K=2) + correction -> +1 tokens (118.1 ms) cpos=199
520
+ [margins] 5.38 8.61 7.02 1.27 4.35 0.80 K=4
521
+ verify 83: accepted 0/3 (K=4) + correction -> +1 tokens (140.1 ms) cpos=200
522
+ [margins] 9.06 3.35 1.36 0.90 1.25 1.25 K=3
523
+ verify 84: accepted 2/2 (K=3) + correction -> +3 tokens (118.2 ms) cpos=203
524
+ [margins] 5.01 0.86 3.03 0.73 0.81 1.39 K=2
525
+ verify 85: accepted 1/1 (K=2) + correction -> +2 tokens (114.4 ms) cpos=205
526
+ [margins] 2.80 0.80 1.58 0.50 0.05 3.80 K=2
527
+ verify 86: accepted 1/1 (K=2) + correction -> +2 tokens (105.4 ms) cpos=207
528
+ [margins] 1.77 3.12 0.60 3.35 3.58 1.28 K=3
529
+ verify 87: accepted 0/2 (K=3) + correction -> +1 tokens (126.8 ms) cpos=208
530
+ [margins] 14.01 1.38 1.41 4.86 1.27 1.32 K=2
531
+ verify 88: accepted 1/1 (K=2) + correction -> +2 tokens (107.8 ms) cpos=210
532
+ [margins] 3.84 5.64 0.38 1.98 3.56 3.26 K=3
533
+ verify 89: accepted 2/2 (K=3) + correction -> +3 tokens (126.7 ms) cpos=213
534
+ [margins] 0.23 2.81 0.82 2.00 1.17 1.31 K=3
535
+ verify 90: accepted 2/2 (K=3) + correction -> +3 tokens (125.8 ms) cpos=216
536
+ [margins] 4.22 6.63 3.23 0.44 2.06 0.45 K=4
537
+ verify 91: accepted 3/3 (K=4) + correction -> +4 tokens (139.5 ms) cpos=220
538
+ [margins] 4.17 4.09 2.36 1.09 4.09 1.08 K=4
539
+ verify 92: accepted 0/3 (K=4) + correction -> +1 tokens (141.7 ms) cpos=221
540
+ [margins] 11.04 2.38 2.16 0.34 3.90 2.81 K=4
541
+ verify 93: accepted 3/3 (K=4) + correction -> +4 tokens (139.7 ms) cpos=225
542
+ [margins] 12.83 6.90 3.22 0.46 0.87 0.90 K=4
543
+ verify 94: accepted 2/3 (K=4) + correction -> +3 tokens (141.9 ms) cpos=228
544
+
545
+ [spec] generated 202 tokens over 94 verifies: mean tokens/verify = 2.15 (block=6, max 7)
546
+ [spec] tokens: 1 16 1 16 1 16 1 16 1 16 1 16 1 16 1 16 1 16 1 16 1 16 1 16 1 16 1 16 1 16 1 16 1 16 1 16 1 16 1 16
547
+ [spec] SPEC-DECODE: 56.7 ms/tok = 17.62 tok/s (vs base M=1 70.7 ms/tok = 14.14 tok/s -> 1.25x)
548
+ [decode] PREFILL: 14 positions in 286.6 ms = 48.8 tok/s (20.475 ms/tok)
549
+ [spec] decoding 200 tokens (block=6, draft passes=1, adaptK=1.50, prompt=4 s=15)...
550
+ [margins] 0.71 7.95 5.55 1.09 5.11 8.00 K=4
551
+ verify 1: accepted 0/3 (K=4) + correction -> +1 tokens (129.1 ms) cpos=15
552
+ [margins] 0.13 0.34 0.30 4.11 0.73 1.89 K=2
553
+ verify 2: accepted 0/1 (K=2) + correction -> +1 tokens (106.2 ms) cpos=16
554
+ [margins] 0.66 0.95 1.73 1.34 0.50 0.55 K=2
555
+ verify 3: accepted 0/1 (K=2) + correction -> +1 tokens (100.6 ms) cpos=17
556
+ [margins] 0.22 0.22 0.76 0.06 1.33 0.08 K=2
557
+ verify 4: accepted 0/1 (K=2) + correction -> +1 tokens (98.3 ms) cpos=18
558
+ [margins] 1.22 0.15 0.28 1.59 0.43 0.76 K=2
559
+ verify 5: accepted 1/1 (K=2) + correction -> +2 tokens (105.1 ms) cpos=20
560
+ [margins] 0.82 1.21 0.17 0.71 3.19 0.23 K=2
561
+ verify 6: accepted 1/1 (K=2) + correction -> +2 tokens (97.2 ms) cpos=22
562
+ [margins] 0.66 0.39 1.00 0.02 0.92 3.85 K=2
563
+ verify 7: accepted 1/1 (K=2) + correction -> +2 tokens (107.2 ms) cpos=24
564
+ [margins] 0.35 1.79 1.24 0.48 4.51 0.72 K=3
565
+ verify 8: accepted 0/2 (K=3) + correction -> +1 tokens (111.9 ms) cpos=25
566
+ [margins] 0.49 0.12 0.99 2.13 0.05 1.98 K=2
567
+ verify 9: accepted 1/1 (K=2) + correction -> +2 tokens (99.4 ms) cpos=27
568
+ [margins] 0.19 0.33 0.00 0.34 1.93 0.86 K=2
569
+ verify 10: accepted 1/1 (K=2) + correction -> +2 tokens (109.9 ms) cpos=29
570
+ [margins] 5.43 3.55 2.31 1.87 1.91 3.53 K=7
571
+ verify 11: accepted 4/6 (K=7) + correction -> +5 tokens (179.7 ms) cpos=34
572
+ [margins] 1.19 1.56 0.90 0.29 1.34 0.77 K=3
573
+ verify 12: accepted 1/2 (K=3) + correction -> +2 tokens (115.7 ms) cpos=36
574
+ [margins] 0.53 10.08 1.12 1.35 0.68 0.30 K=3
575
+ verify 13: accepted 0/2 (K=3) + correction -> +1 tokens (109.7 ms) cpos=37
576
+ [margins] 0.53 3.74 0.91 0.46 0.94 0.29 K=3
577
+ verify 14: accepted 0/2 (K=3) + correction -> +1 tokens (117.0 ms) cpos=38
578
+ [margins] 0.16 0.44 2.50 0.72 1.16 0.42 K=2
579
+ verify 15: accepted 0/1 (K=2) + correction -> +1 tokens (107.1 ms) cpos=39
580
+ [margins] 4.15 1.43 0.04 0.47 0.58 0.65 K=2
581
+ verify 16: accepted 0/1 (K=2) + correction -> +1 tokens (105.6 ms) cpos=40
582
+ [margins] 3.50 3.15 1.30 1.43 1.57 0.25 K=3
583
+ verify 17: accepted 1/2 (K=3) + correction -> +2 tokens (113.2 ms) cpos=42
584
+ [margins] 0.38 1.66 0.08 0.79 0.12 4.42 K=3
585
+ verify 18: accepted 1/2 (K=3) + correction -> +2 tokens (120.6 ms) cpos=44
586
+ [margins] 0.25 4.43 0.76 1.19 1.81 2.01 K=3
587
+ verify 19: accepted 2/2 (K=3) + correction -> +3 tokens (113.4 ms) cpos=47
588
+ [margins] 3.71 7.65 3.51 1.12 2.79 1.04 K=4
589
+ verify 20: accepted 3/3 (K=4) + correction -> +4 tokens (132.1 ms) cpos=51
590
+ [margins] 2.72 0.48 1.47 0.59 1.34 0.22 K=2
591
+ verify 21: accepted 0/1 (K=2) + correction -> +1 tokens (108.1 ms) cpos=52
592
+ [margins] 1.69 2.39 0.73 1.72 2.98 0.74 K=3
593
+ verify 22: accepted 0/2 (K=3) + correction -> +1 tokens (111.9 ms) cpos=53
594
+ [margins] 8.24 3.04 0.36 1.04 1.24 1.23 K=3
595
+ verify 23: accepted 2/2 (K=3) + correction -> +3 tokens (123.2 ms) cpos=56
596
+ [margins] 4.56 1.90 4.09 5.18 0.35 0.34 K=5
597
+ verify 24: accepted 4/4 (K=5) + correction -> +5 tokens (142.6 ms) cpos=61
598
+ [margins] 1.81 2.98 1.91 1.26 1.30 0.67 K=4
599
+ verify 25: accepted 0/3 (K=4) + correction -> +1 tokens (131.9 ms) cpos=62
600
+ [margins] 1.94 1.41 2.38 0.06 2.91 3.27 K=2
601
+ verify 26: accepted 0/1 (K=2) + correction -> +1 tokens (106.2 ms) cpos=63
602
+ [margins] 0.34 1.87 0.88 0.23 1.11 1.47 K=3
603
+ verify 27: accepted 2/2 (K=3) + correction -> +3 tokens (121.6 ms) cpos=66
604
+ [margins] 0.29 4.84 1.02 0.21 0.28 1.28 K=3
605
+ verify 28: accepted 0/2 (K=3) + correction -> +1 tokens (119.6 ms) cpos=67
606
+ [margins] 0.45 0.20 0.29 1.68 1.93 0.44 K=2
607
+ verify 29: accepted 1/1 (K=2) + correction -> +2 tokens (110.0 ms) cpos=69
608
+ [margins] 0.03 3.74 1.45 0.54 0.07 0.20 K=3
609
+ verify 30: accepted 1/2 (K=3) + correction -> +2 tokens (120.3 ms) cpos=71
610
+ [margins] 0.37 0.04 0.06 0.47 2.22 0.21 K=2
611
+ verify 31: accepted 1/1 (K=2) + correction -> +2 tokens (110.0 ms) cpos=73
612
+ [margins] 5.62 0.45 0.28 0.21 0.16 1.25 K=2
613
+ verify 32: accepted 1/1 (K=2) + correction -> +2 tokens (101.9 ms) cpos=75
614
+ [margins] 1.52 1.20 1.65 1.81 0.05 2.61 K=2
615
+ verify 33: accepted 1/1 (K=2) + correction -> +2 tokens (108.5 ms) cpos=77
616
+ [margins] 5.27 1.16 0.84 0.16 1.09 0.64 K=2
617
+ verify 34: accepted 1/1 (K=2) + correction -> +2 tokens (101.8 ms) cpos=79
618
+ [margins] 3.17 2.30 0.91 0.74 0.05 8.07 K=3
619
+ verify 35: accepted 2/2 (K=3) + correction -> +3 tokens (123.0 ms) cpos=82
620
+ [margins] 2.05 3.57 10.22 0.49 0.30 9.01 K=4
621
+ verify 36: accepted 3/3 (K=4) + correction -> +4 tokens (140.9 ms) cpos=86
622
+ [margins] 7.42 6.71 1.92 2.58 0.19 2.24 K=5
623
+ verify 37: accepted 2/4 (K=5) + correction -> +3 tokens (136.6 ms) cpos=89
624
+ [margins] 7.60 5.48 0.39 0.27 0.41 3.22 K=3
625
+ verify 38: accepted 1/2 (K=3) + correction -> +2 tokens (119.7 ms) cpos=91
626
+ [margins] 6.25 3.58 1.11 0.08 2.27 5.27 K=3
627
+ verify 39: accepted 1/2 (K=3) + correction -> +2 tokens (118.7 ms) cpos=93
628
+ [margins] 4.81 0.77 5.52 4.62 1.00 2.60 K=2
629
+ verify 40: accepted 1/1 (K=2) + correction -> +2 tokens (105.8 ms) cpos=95
630
+ [margins] 4.30 1.55 6.37 4.13 1.84 4.26 K=7
631
+ verify 41: accepted 1/6 (K=7) + correction -> +2 tokens (183.1 ms) cpos=97
632
+ [margins] 2.35 0.52 7.68 3.69 1.25 5.55 K=2
633
+ verify 42: accepted 1/1 (K=2) + correction -> +2 tokens (106.5 ms) cpos=99
634
+ [margins] 1.90 0.35 7.86 4.15 0.92 5.43 K=2
635
+ verify 43: accepted 1/1 (K=2) + correction -> +2 tokens (113.0 ms) cpos=101
636
+ [margins] 2.88 0.51 8.86 3.82 1.07 6.59 K=2
637
+ verify 44: accepted 1/1 (K=2) + correction -> +2 tokens (106.2 ms) cpos=103
638
+ [margins] 2.92 0.49 10.00 4.47 0.50 6.45 K=2
639
+ verify 45: accepted 1/1 (K=2) + correction -> +2 tokens (113.1 ms) cpos=105
640
+ [margins] 2.71 0.50 9.85 4.05 0.50 6.46 K=2
641
+ verify 46: accepted 1/1 (K=2) + correction -> +2 tokens (106.7 ms) cpos=107
642
+ [margins] 4.17 0.46 9.11 4.26 0.57 6.80 K=2
643
+ verify 47: accepted 1/1 (K=2) + correction -> +2 tokens (113.6 ms) cpos=109
644
+ [margins] 2.81 0.11 7.95 3.70 0.06 6.28 K=2
645
+ verify 48: accepted 1/1 (K=2) + correction -> +2 tokens (106.9 ms) cpos=111
646
+ [margins] 2.68 0.62 8.78 3.42 0.07 6.41 K=2
647
+ verify 49: accepted 1/1 (K=2) + correction -> +2 tokens (113.8 ms) cpos=113
648
+ [margins] 2.48 0.45 8.92 3.44 0.34 6.54 K=2
649
+ verify 50: accepted 1/1 (K=2) + correction -> +2 tokens (107.0 ms) cpos=115
650
+ [margins] 2.65 0.02 8.69 3.67 0.21 6.72 K=2
651
+ verify 51: accepted 1/1 (K=2) + correction -> +2 tokens (114.2 ms) cpos=117
652
+ [margins] 2.91 0.11 8.34 3.90 0.74 6.23 K=2
653
+ verify 52: accepted 1/1 (K=2) + correction -> +2 tokens (107.9 ms) cpos=119
654
+ [margins] 2.50 0.31 8.17 3.23 0.27 6.20 K=2
655
+ verify 53: accepted 1/1 (K=2) + correction -> +2 tokens (115.3 ms) cpos=121
656
+ [margins] 4.05 0.24 8.41 3.18 0.38 6.61 K=2
657
+ verify 54: accepted 1/1 (K=2) + correction -> +2 tokens (107.3 ms) cpos=123
658
+ [margins] 3.72 0.14 9.06 3.69 0.70 6.20 K=2
659
+ verify 55: accepted 1/1 (K=2) + correction -> +2 tokens (114.4 ms) cpos=125
660
+ [margins] 4.07 0.02 7.60 3.38 0.02 6.32 K=2
661
+ verify 56: accepted 1/1 (K=2) + correction -> +2 tokens (108.1 ms) cpos=127
662
+ [margins] 4.03 0.03 6.41 3.53 0.57 6.06 K=2
663
+ verify 57: accepted 1/1 (K=2) + correction -> +2 tokens (155.8 ms) cpos=129
664
+ [margins] 4.63 0.49 8.69 4.02 0.28 5.55 K=2
665
+ verify 58: accepted 1/1 (K=2) + correction -> +2 tokens (108.3 ms) cpos=131
666
+ [margins] 3.98 0.20 8.71 4.27 0.91 5.44 K=2
667
+ verify 59: accepted 1/1 (K=2) + correction -> +2 tokens (115.4 ms) cpos=133
668
+ [margins] 3.75 0.40 9.01 2.98 0.05 5.92 K=2
669
+ verify 60: accepted 1/1 (K=2) + correction -> +2 tokens (107.6 ms) cpos=135
670
+ [margins] 4.36 0.98 8.78 2.97 0.13 6.78 K=2
671
+ verify 61: accepted 1/1 (K=2) + correction -> +2 tokens (116.1 ms) cpos=137
672
+ [margins] 4.71 0.65 8.50 3.58 0.55 5.51 K=2
673
+ verify 62: accepted 1/1 (K=2) + correction -> +2 tokens (108.7 ms) cpos=139
674
+ [margins] 4.92 0.20 5.94 3.66 0.04 5.86 K=2
675
+ verify 63: accepted 1/1 (K=2) + correction -> +2 tokens (116.1 ms) cpos=141
676
+ [margins] 4.68 0.18 8.54 3.82 0.06 6.77 K=2
677
+ verify 64: accepted 1/1 (K=2) + correction -> +2 tokens (108.2 ms) cpos=143
678
+ [margins] 4.91 0.00 7.45 4.12 0.16 5.63 K=2
679
+ verify 65: accepted 1/1 (K=2) + correction -> +2 tokens (115.7 ms) cpos=145
680
+ [margins] 4.62 0.73 8.46 3.71 0.13 5.40 K=2
681
+ verify 66: accepted 1/1 (K=2) + correction -> +2 tokens (108.5 ms) cpos=147
682
+ [margins] 4.86 0.76 8.13 3.50 0.23 5.40 K=2
683
+ verify 67: accepted 1/1 (K=2) + correction -> +2 tokens (115.4 ms) cpos=149
684
+ [margins] 3.09 0.03 6.63 3.31 1.01 6.13 K=2
685
+ verify 68: accepted 1/1 (K=2) + correction -> +2 tokens (108.4 ms) cpos=151
686
+ [margins] 3.76 0.25 8.92 3.19 0.75 5.68 K=2
687
+ verify 69: accepted 1/1 (K=2) + correction -> +2 tokens (115.2 ms) cpos=153
688
+ [margins] 3.90 0.80 9.28 2.56 0.20 4.76 K=2
689
+ verify 70: accepted 1/1 (K=2) + correction -> +2 tokens (108.7 ms) cpos=155
690
+ [margins] 3.97 0.98 9.08 2.85 0.90 5.94 K=2
691
+ verify 71: accepted 0/1 (K=2) + correction -> +1 tokens (115.2 ms) cpos=156
692
+ [margins] 0.44 3.62 0.06 6.81 2.63 0.48 K=3
693
+ verify 72: accepted 1/2 (K=3) + correction -> +2 tokens (120.9 ms) cpos=158
694
+ [margins] 3.08 0.82 1.45 2.12 1.50 1.16 K=2
695
+ verify 73: accepted 0/1 (K=2) + correction -> +1 tokens (112.9 ms) cpos=159
696
+ [margins] 7.90 3.46 0.27 0.45 0.13 0.86 K=3
697
+ verify 74: accepted 2/2 (K=3) + correction -> +3 tokens (124.7 ms) cpos=162
698
+ [margins] 5.17 0.33 2.19 0.38 0.45 6.72 K=2
699
+ verify 75: accepted 1/1 (K=2) + correction -> +2 tokens (113.5 ms) cpos=164
700
+ [margins] 7.94 5.28 1.71 3.11 0.96 1.48 K=5
701
+ verify 76: accepted 2/4 (K=5) + correction -> +3 tokens (148.9 ms) cpos=167
702
+ [margins] 12.10 4.30 8.73 8.47 5.08 0.52 K=6
703
+ verify 77: accepted 5/5 (K=6) + correction -> +6 tokens (175.0 ms) cpos=173
704
+ [margins] 5.15 0.27 0.14 1.49 0.38 5.69 K=2
705
+ verify 78: accepted 1/1 (K=2) + correction -> +2 tokens (105.4 ms) cpos=175
706
+ [margins] 7.71 9.20 5.06 6.81 0.10 0.55 K=5
707
+ verify 79: accepted 4/4 (K=5) + correction -> +5 tokens (159.1 ms) cpos=180
708
+ [margins] 12.81 11.78 12.58 3.95 8.01 5.37 K=7
709
+ verify 80: accepted 6/6 (K=7) + correction -> +7 tokens (190.4 ms) cpos=187
710
+ [margins] 17.53 5.39 4.81 2.17 1.10 5.37 K=5
711
+ verify 81: accepted 4/4 (K=5) + correction -> +5 tokens (159.0 ms) cpos=192
712
+ [margins] 10.94 8.39 1.47 1.12 11.07 13.10 K=3
713
+ verify 82: accepted 2/2 (K=3) + correction -> +3 tokens (119.8 ms) cpos=195
714
+ [margins] 17.36 9.75 11.17 10.98 6.52 5.97 K=7
715
+ verify 83: accepted 6/6 (K=7) + correction -> +7 tokens (199.0 ms) cpos=202
716
+ [margins] 12.79 14.01 16.54 5.84 5.64 8.79 K=7
717
+ verify 84: accepted 6/6 (K=7) + correction -> +7 tokens (197.0 ms) cpos=209
718
+ [margins] 9.79 10.17 11.59 8.75 2.23 14.52 K=7
719
+ verify 85: accepted 2/6 (K=7) + correction -> +3 tokens (195.2 ms) cpos=212
720
+ [margins] 16.18 12.13 12.06 10.09 6.11 0.17 K=6
721
+ verify 86: accepted 5/5 (K=6) + correction -> +6 tokens (170.2 ms) cpos=218
722
+
723
+ [spec] generated 204 tokens over 86 verifies: mean tokens/verify = 2.37 (block=6, max 7)
724
+ [spec] tokens: 5715 14 295 834 10175 14 8470 3939 436 344 2239 362 4197 339 35 126664 23809 16765 270 3077 305 1990 23714 538 9479 21121 35977 2184 47270 34788 14 832 851 588 366 99782 1082 20351 14319 35977
725
+ [spec] SPEC-DECODE: 51.1 ms/tok = 19.59 tok/s (vs base M=1 70.7 ms/tok = 14.14 tok/s -> 1.39x)
726
+ [decode] PREFILL: 82 positions in 1369.1 ms = 59.9 tok/s (16.696 ms/tok)
727
+ [spec] decoding 200 tokens (block=6, draft passes=1, adaptK=1.50, prompt=5 s=83)...
728
+ [margins] 8.27 2.69 1.01 0.44 3.13 2.25 K=3
729
+ verify 1: accepted 1/2 (K=3) + correction -> +2 tokens (125.3 ms) cpos=84
730
+ [margins] 0.60 1.88 3.49 1.04 0.85 1.45 K=4
731
+ verify 2: accepted 0/3 (K=4) + correction -> +1 tokens (133.3 ms) cpos=85
732
+ [margins] 1.20 0.47 0.27 0.44 0.56 0.96 K=2
733
+ verify 3: accepted 0/1 (K=2) + correction -> +1 tokens (102.0 ms) cpos=86
734
+ [margins] 6.94 6.58 2.69 5.41 0.46 3.99 K=5
735
+ verify 4: accepted 4/4 (K=5) + correction -> +5 tokens (147.1 ms) cpos=91
736
+ [margins] 12.51 7.41 5.73 4.48 3.71 0.71 K=6
737
+ verify 5: accepted 5/5 (K=6) + correction -> +6 tokens (171.6 ms) cpos=97
738
+ [margins] 13.65 3.96 1.24 0.98 0.10 0.97 K=3
739
+ verify 6: accepted 2/2 (K=3) + correction -> +3 tokens (123.7 ms) cpos=100
740
+ [margins] 0.91 1.59 2.97 0.89 1.69 2.38 K=4
741
+ verify 7: accepted 2/3 (K=4) + correction -> +3 tokens (137.2 ms) cpos=103
742
+ [margins] 3.56 1.79 2.11 1.61 1.89 3.37 K=7
743
+ verify 8: accepted 1/6 (K=7) + correction -> +2 tokens (190.6 ms) cpos=105
744
+ [margins] 10.52 7.54 8.78 6.13 1.37 6.15 K=5
745
+ verify 9: accepted 4/4 (K=5) + correction -> +5 tokens (149.1 ms) cpos=110
746
+ [margins] 13.51 8.40 8.65 7.04 12.45 1.06 K=6
747
+ verify 10: accepted 5/5 (K=6) + correction -> +6 tokens (174.3 ms) cpos=116
748
+ [margins] 14.06 14.30 7.08 1.18 2.60 0.80 K=4
749
+ verify 11: accepted 3/3 (K=4) + correction -> +4 tokens (138.5 ms) cpos=120
750
+ [margins] 2.77 0.49 5.02 0.24 5.77 0.07 K=2
751
+ verify 12: accepted 1/1 (K=2) + correction -> +2 tokens (108.5 ms) cpos=122
752
+ [margins] 7.62 3.63 7.09 1.80 4.34 4.20 K=7
753
+ verify 13: accepted 3/6 (K=7) + correction -> +4 tokens (236.1 ms) cpos=126
754
+ [margins] 11.55 8.31 4.07 11.32 5.81 0.09 K=6
755
+ verify 14: accepted 5/5 (K=6) + correction -> +6 tokens (216.1 ms) cpos=132
756
+ [margins] 11.48 15.02 2.02 9.69 9.32 12.52 K=7
757
+ verify 15: accepted 2/6 (K=7) + correction -> +3 tokens (187.5 ms) cpos=135
758
+ [margins] 14.78 15.02 10.73 8.07 0.45 3.51 K=5
759
+ verify 16: accepted 4/4 (K=5) + correction -> +5 tokens (157.8 ms) cpos=140
760
+ [margins] 4.63 1.47 4.06 4.02 5.77 2.79 K=2
761
+ verify 17: accepted 0/1 (K=2) + correction -> +1 tokens (105.4 ms) cpos=141
762
+ [margins] 10.00 5.75 2.40 2.67 0.07 0.17 K=5
763
+ verify 18: accepted 2/4 (K=5) + correction -> +3 tokens (146.5 ms) cpos=144
764
+ [margins] 10.05 8.72 6.91 4.45 1.91 6.33 K=7
765
+ verify 19: accepted 4/6 (K=7) + correction -> +5 tokens (189.0 ms) cpos=149
766
+ [margins] 13.12 10.60 3.88 5.98 11.67 0.12 K=6
767
+ verify 20: accepted 5/5 (K=6) + correction -> +6 tokens (167.9 ms) cpos=155
768
+ [margins] 14.43 12.11 9.99 8.51 1.27 2.98 K=5
769
+ verify 21: accepted 4/4 (K=5) + correction -> +5 tokens (157.7 ms) cpos=160
770
+ [margins] 6.29 1.25 8.66 6.13 3.43 7.27 K=2
771
+ verify 22: accepted 0/1 (K=2) + correction -> +1 tokens (105.9 ms) cpos=161
772
+ [margins] 12.81 6.31 3.97 0.35 3.02 0.64 K=4
773
+ verify 23: accepted 2/3 (K=4) + correction -> +3 tokens (135.7 ms) cpos=164
774
+ [margins] 10.14 10.41 6.96 9.22 4.33 4.85 K=7
775
+ verify 24: accepted 6/6 (K=7) + correction -> +7 tokens (189.1 ms) cpos=171
776
+ [margins] 2.78 9.09 8.28 2.78 7.69 9.61 K=7
777
+ verify 25: accepted 3/6 (K=7) + correction -> +4 tokens (197.0 ms) cpos=175
778
+ [margins] 16.92 12.72 11.54 7.25 0.37 0.41 K=5
779
+ verify 26: accepted 4/4 (K=5) + correction -> +5 tokens (158.4 ms) cpos=180
780
+ [margins] 16.65 15.12 3.57 1.61 7.90 2.87 K=7
781
+ verify 27: accepted 5/6 (K=7) + correction -> +6 tokens (187.8 ms) cpos=186
782
+ [margins] 6.96 9.37 2.20 8.09 0.31 10.54 K=5
783
+ verify 28: accepted 4/4 (K=5) + correction -> +5 tokens (153.5 ms) cpos=191
784
+ [margins] 9.38 11.35 9.26 4.88 12.09 5.80 K=7
785
+ verify 29: accepted 3/6 (K=7) + correction -> +4 tokens (197.1 ms) cpos=195
786
+ [margins] 14.67 14.08 11.80 9.36 0.76 4.63 K=5
787
+ verify 30: accepted 4/4 (K=5) + correction -> +5 tokens (158.6 ms) cpos=200
788
+ [margins] 18.25 16.10 4.09 1.02 6.39 2.91 K=4
789
+ verify 31: accepted 3/3 (K=4) + correction -> +4 tokens (137.2 ms) cpos=204
790
+ [margins] 9.95 4.93 5.51 6.76 0.78 9.45 K=5
791
+ verify 32: accepted 4/4 (K=5) + correction -> +5 tokens (150.9 ms) cpos=209
792
+ [margins] 14.58 13.05 5.52 9.32 10.76 0.91 K=6
793
+ verify 33: accepted 5/5 (K=6) + correction -> +6 tokens (169.1 ms) cpos=215
794
+ [margins] 15.31 14.21 9.74 10.92 5.20 0.62 K=6
795
+ verify 34: accepted 5/5 (K=6) + correction -> +6 tokens (177.1 ms) cpos=221
796
+ [margins] 13.97 8.26 1.98 8.33 1.70 9.53 K=7
797
+ verify 35: accepted 4/6 (K=7) + correction -> +5 tokens (197.0 ms) cpos=226
798
+ [margins] 9.51 3.22 1.68 12.05 0.27 0.11 K=5
799
+ verify 36: accepted 4/4 (K=5) + correction -> +5 tokens (154.0 ms) cpos=231
800
+ [margins] 10.22 8.37 7.64 1.03 9.29 1.52 K=4
801
+ verify 37: accepted 3/3 (K=4) + correction -> +4 tokens (144.0 ms) cpos=235
802
+ [margins] 10.12 11.47 6.82 7.32 2.83 4.58 K=7
803
+ verify 38: accepted 5/6 (K=7) + correction -> +6 tokens (198.7 ms) cpos=241
804
+ [margins] 15.93 9.34 6.25 7.67 0.39 2.56 K=5
805
+ verify 39: accepted 4/4 (K=5) + correction -> +5 tokens (154.4 ms) cpos=246
806
+ [margins] 10.08 2.82 1.16 5.44 13.26 4.08 K=3
807
+ verify 40: accepted 2/2 (K=3) + correction -> +3 tokens (130.5 ms) cpos=249
808
+ [margins] 14.83 12.37 4.12 10.58 12.03 1.15 K=6
809
+ verify 41: accepted 5/5 (K=6) + correction -> +6 tokens (169.6 ms) cpos=255
810
+ [margins] 15.51 13.94 11.00 10.74 6.56 1.99 K=7
811
+ verify 42: accepted 6/6 (K=7) + correction -> +7 tokens (240.8 ms) cpos=262
812
+ [margins] 11.50 4.11 5.06 1.44 4.65 6.49 K=4
813
+ verify 43: accepted 3/3 (K=4) + correction -> +4 tokens (143.1 ms) cpos=266
814
+ [margins] 10.93 11.93 1.71 7.16 10.75 9.46 K=7
815
+ verify 44: accepted 2/6 (K=7) + correction -> +3 tokens (199.0 ms) cpos=269
816
+ [margins] 14.90 12.49 4.78 9.16 10.47 1.40 K=6
817
+ verify 45: accepted 5/5 (K=6) + correction -> +6 tokens (173.8 ms) cpos=275
818
+ [margins] 13.48 14.41 12.72 10.42 8.45 4.58 K=7
819
+ verify 46: accepted 6/6 (K=7) + correction -> +7 tokens (200.1 ms) cpos=282
820
+
821
+ [spec] generated 200 tokens over 46 verifies: mean tokens/verify = 4.35 (block=6, max 7)
822
+ [spec] tokens: 1 50 201 671 15020 4181 362 2775 14817 14922 344 223 25 22765 7975 55 15 3180 16 1 50 201 671 15020 4181 362 2775 14817 14922 344 223 25 22765 7975 55 15 3180 16 1 50
823
+ [spec] SPEC-DECODE: 37.2 ms/tok = 26.89 tok/s (vs base M=1 70.7 ms/tok = 14.14 tok/s -> 1.90x)
824
+ [decode] PREFILL: 14 positions in 293.2 ms = 47.8 tok/s (20.941 ms/tok)
825
+ [spec] decoding 200 tokens (block=6, draft passes=1, adaptK=1.50, prompt=6 s=15)...
826
+ [margins] 0.61 1.21 0.94 1.25 0.31 0.20 K=2
827
+ verify 1: accepted 0/1 (K=2) + correction -> +1 tokens (106.3 ms) cpos=15
828
+ [margins] 0.42 0.56 0.33 0.16 0.29 0.09 K=2
829
+ verify 2: accepted 0/1 (K=2) + correction -> +1 tokens (109.5 ms) cpos=16
830
+ [margins] 2.97 0.60 1.30 0.44 1.78 0.08 K=2
831
+ verify 3: accepted 1/1 (K=2) + correction -> +2 tokens (101.7 ms) cpos=18
832
+ [margins] 2.91 0.91 0.34 0.16 0.67 1.51 K=2
833
+ verify 4: accepted 1/1 (K=2) + correction -> +2 tokens (108.3 ms) cpos=20
834
+ [margins] 3.22 0.93 3.87 3.58 2.41 1.89 K=2
835
+ verify 5: accepted 1/1 (K=2) + correction -> +2 tokens (100.9 ms) cpos=22
836
+ [margins] 3.27 2.33 4.08 3.50 3.03 2.92 K=7
837
+ verify 6: accepted 1/6 (K=7) + correction -> +2 tokens (173.2 ms) cpos=24
838
+ [margins] 2.82 2.76 5.09 2.88 1.65 3.87 K=7
839
+ verify 7: accepted 1/6 (K=7) + correction -> +2 tokens (165.6 ms) cpos=26
840
+ [margins] 3.36 1.30 6.72 3.07 0.91 5.14 K=2
841
+ verify 8: accepted 1/1 (K=2) + correction -> +2 tokens (108.2 ms) cpos=28
842
+ [margins] 3.15 0.06 7.82 3.39 0.69 5.90 K=2
843
+ verify 9: accepted 1/1 (K=2) + correction -> +2 tokens (101.5 ms) cpos=30
844
+ [margins] 2.66 0.32 8.49 3.68 0.27 6.33 K=2
845
+ verify 10: accepted 0/1 (K=2) + correction -> +1 tokens (107.9 ms) cpos=31
846
+ [margins] 0.02 7.93 2.81 0.61 6.44 3.53 K=4
847
+ verify 11: accepted 0/3 (K=4) + correction -> +1 tokens (131.5 ms) cpos=32
848
+ [margins] 7.64 2.17 0.23 7.57 3.28 1.23 K=3
849
+ verify 12: accepted 1/2 (K=3) + correction -> +2 tokens (112.2 ms) cpos=34
850
+ [margins] 3.71 5.93 0.48 0.12 4.66 1.84 K=3
851
+ verify 13: accepted 2/2 (K=3) + correction -> +3 tokens (116.7 ms) cpos=37
852
+ [margins] 11.37 6.73 3.51 8.07 6.60 1.83 K=7
853
+ verify 14: accepted 6/6 (K=7) + correction -> +7 tokens (175.4 ms) cpos=44
854
+ [margins] 9.09 11.16 13.69 9.04 9.44 10.39 K=7
855
+ verify 15: accepted 6/6 (K=7) + correction -> +7 tokens (170.0 ms) cpos=51
856
+ [margins] 13.06 13.76 8.63 10.92 11.16 7.97 K=7
857
+ verify 16: accepted 6/6 (K=7) + correction -> +7 tokens (179.5 ms) cpos=58
858
+ [margins] 17.60 9.88 11.05 3.34 10.34 6.89 K=7
859
+ verify 17: accepted 6/6 (K=7) + correction -> +7 tokens (178.9 ms) cpos=65
860
+ [margins] 8.54 12.81 11.92 9.08 9.82 8.83 K=7
861
+ verify 18: accepted 6/6 (K=7) + correction -> +7 tokens (182.9 ms) cpos=72
862
+ [margins] 13.20 14.25 8.61 10.69 11.45 8.63 K=7
863
+ verify 19: accepted 6/6 (K=7) + correction -> +7 tokens (176.5 ms) cpos=79
864
+ [margins] 18.10 10.92 11.36 12.91 11.24 8.70 K=7
865
+ verify 20: accepted 6/6 (K=7) + correction -> +7 tokens (183.5 ms) cpos=86
866
+ [margins] 10.32 14.60 12.72 10.19 9.49 10.79 K=7
867
+ verify 21: accepted 6/6 (K=7) + correction -> +7 tokens (184.8 ms) cpos=93
868
+ [margins] 12.93 13.86 9.56 11.56 11.00 8.55 K=7
869
+ verify 22: accepted 6/6 (K=7) + correction -> +7 tokens (187.7 ms) cpos=100
870
+ [margins] 17.85 9.89 12.47 10.91 9.97 8.62 K=7
871
+ verify 23: accepted 6/6 (K=7) + correction -> +7 tokens (180.2 ms) cpos=107
872
+ [margins] 10.04 13.81 14.18 11.22 9.61 13.51 K=7
873
+ verify 24: accepted 6/6 (K=7) + correction -> +7 tokens (189.3 ms) cpos=114
874
+ [margins] 12.19 15.69 9.81 11.98 10.78 8.20 K=7
875
+ verify 25: accepted 6/6 (K=7) + correction -> +7 tokens (188.0 ms) cpos=121
876
+ [margins] 17.25 11.79 13.34 9.17 11.99 9.28 K=7
877
+ verify 26: accepted 6/6 (K=7) + correction -> +7 tokens (231.7 ms) cpos=128
878
+ [margins] 10.04 13.85 13.40 9.33 9.20 12.48 K=7
879
+ verify 27: accepted 6/6 (K=7) + correction -> +7 tokens (184.1 ms) cpos=135
880
+ [margins] 15.34 14.45 10.08 12.14 10.13 9.21 K=7
881
+ verify 28: accepted 6/6 (K=7) + correction -> +7 tokens (188.5 ms) cpos=142
882
+ [margins] 16.31 11.29 12.97 11.39 10.94 9.13 K=7
883
+ verify 29: accepted 6/6 (K=7) + correction -> +7 tokens (189.1 ms) cpos=149
884
+ [margins] 8.95 13.66 14.29 10.01 10.61 12.76 K=7
885
+ verify 30: accepted 6/6 (K=7) + correction -> +7 tokens (188.8 ms) cpos=156
886
+ [margins] 13.15 15.36 9.77 11.69 13.34 7.32 K=7
887
+ verify 31: accepted 6/6 (K=7) + correction -> +7 tokens (186.1 ms) cpos=163
888
+ [margins] 19.41 10.72 12.37 12.71 9.01 9.33 K=7
889
+ verify 32: accepted 6/6 (K=7) + correction -> +7 tokens (189.7 ms) cpos=170
890
+ [margins] 9.89 12.61 12.30 9.64 8.46 9.97 K=7
891
+ verify 33: accepted 6/6 (K=7) + correction -> +7 tokens (191.3 ms) cpos=177
892
+ [margins] 13.25 14.62 9.89 11.60 10.70 8.27 K=7
893
+ verify 34: accepted 6/6 (K=7) + correction -> +7 tokens (191.9 ms) cpos=184
894
+ [margins] 18.69 11.55 13.31 6.82 9.55 8.03 K=7
895
+ verify 35: accepted 6/6 (K=7) + correction -> +7 tokens (187.2 ms) cpos=191
896
+ [margins] 9.82 14.17 12.45 9.31 8.87 13.35 K=7
897
+ verify 36: accepted 6/6 (K=7) + correction -> +7 tokens (192.1 ms) cpos=198
898
+ [margins] 14.54 15.18 10.39 11.39 12.95 9.23 K=7
899
+ verify 37: accepted 6/6 (K=7) + correction -> +7 tokens (194.1 ms) cpos=205
900
+ [margins] 18.45 11.01 14.08 9.26 11.64 8.94 K=7
901
+ verify 38: accepted 6/6 (K=7) + correction -> +7 tokens (193.5 ms) cpos=212
902
+ [margins] 10.02 13.64 17.16 10.35 10.94 13.77 K=7
903
+ verify 39: accepted 6/6 (K=7) + correction -> +7 tokens (183.9 ms) cpos=219
904
+
905
+ [spec] generated 205 tokens over 39 verifies: mean tokens/verify = 5.26 (block=6, max 7)
906
+ [spec] tokens: 1 16 1 16 1 16 1 16 1 16 1 16 1 16 1 16 403 31310 16 403 31310 16 403 31310 16 403 31310 16 403 31310 16 403 31310 16 403 31310 16 403 31310 16
907
+ [spec] SPEC-DECODE: 30.9 ms/tok = 32.35 tok/s (vs base M=1 70.7 ms/tok = 14.14 tok/s -> 2.29x)
908
+ [decode] PREFILL: 37 positions in 666.6 ms = 55.5 tok/s (18.016 ms/tok)
909
+ [spec] decoding 200 tokens (block=6, draft passes=1, adaptK=1.50, prompt=7 s=38)...
910
+ [margins] 0.06 2.31 3.48 0.24 0.25 0.66 K=4
911
+ verify 1: accepted 1/3 (K=4) + correction -> +2 tokens (130.6 ms) cpos=39
912
+ [margins] 2.09 0.72 5.03 0.42 1.31 0.82 K=2
913
+ verify 2: accepted 1/1 (K=2) + correction -> +2 tokens (105.9 ms) cpos=41
914
+ [margins] 4.41 0.11 0.30 0.38 0.05 0.18 K=2
915
+ verify 3: accepted 1/1 (K=2) + correction -> +2 tokens (98.6 ms) cpos=43
916
+ [margins] 2.11 2.49 0.55 0.18 0.26 0.20 K=3
917
+ verify 4: accepted 0/2 (K=3) + correction -> +1 tokens (118.0 ms) cpos=44
918
+ [margins] 0.09 0.70 1.02 0.68 0.24 0.34 K=2
919
+ verify 5: accepted 0/1 (K=2) + correction -> +1 tokens (101.1 ms) cpos=45
920
+ [margins] 0.01 0.11 0.71 1.10 0.29 0.02 K=2
921
+ verify 6: accepted 0/1 (K=2) + correction -> +1 tokens (100.1 ms) cpos=46
922
+ [margins] 1.26 1.98 2.03 0.46 1.02 0.12 K=4
923
+ verify 7: accepted 0/3 (K=4) + correction -> +1 tokens (132.2 ms) cpos=47
924
+ [margins] 1.79 1.58 0.76 0.64 1.11 1.82 K=3
925
+ verify 8: accepted 1/2 (K=3) + correction -> +2 tokens (119.8 ms) cpos=49
926
+ [margins] 9.24 6.19 4.86 3.98 0.23 0.72 K=5
927
+ verify 9: accepted 4/4 (K=5) + correction -> +5 tokens (144.5 ms) cpos=54
928
+ [margins] 0.87 1.25 0.55 2.37 0.84 0.76 K=2
929
+ verify 10: accepted 0/1 (K=2) + correction -> +1 tokens (109.1 ms) cpos=55
930
+ [margins] 1.94 1.05 0.88 2.78 1.83 2.07 K=2
931
+ verify 11: accepted 1/1 (K=2) + correction -> +2 tokens (108.6 ms) cpos=57
932
+ [margins] 3.24 0.46 0.03 2.79 1.71 0.55 K=2
933
+ verify 12: accepted 1/1 (K=2) + correction -> +2 tokens (101.4 ms) cpos=59
934
+ [margins] 4.99 1.74 2.22 0.00 0.93 0.22 K=4
935
+ verify 13: accepted 3/3 (K=4) + correction -> +4 tokens (133.7 ms) cpos=63
936
+ [margins] 6.97 3.60 5.73 2.33 2.57 2.10 K=7
937
+ verify 14: accepted 4/6 (K=7) + correction -> +5 tokens (185.1 ms) cpos=68
938
+ [margins] 6.37 2.74 1.10 1.82 0.66 8.00 K=3
939
+ verify 15: accepted 2/2 (K=3) + correction -> +3 tokens (112.3 ms) cpos=71
940
+ [margins] 3.21 1.90 4.10 0.25 0.55 0.94 K=4
941
+ verify 16: accepted 3/3 (K=4) + correction -> +4 tokens (131.5 ms) cpos=75
942
+ [margins] 2.00 9.84 0.14 3.80 0.40 3.72 K=3
943
+ verify 17: accepted 2/2 (K=3) + correction -> +3 tokens (118.9 ms) cpos=78
944
+ [margins] 9.60 8.05 6.10 9.16 8.75 4.35 K=7
945
+ verify 18: accepted 6/6 (K=7) + correction -> +7 tokens (186.1 ms) cpos=85
946
+ [margins] 12.38 9.31 8.17 6.14 2.42 6.14 K=7
947
+ verify 19: accepted 6/6 (K=7) + correction -> +7 tokens (189.1 ms) cpos=92
948
+ [margins] 2.24 8.81 4.44 0.09 3.51 0.11 K=4
949
+ verify 20: accepted 3/3 (K=4) + correction -> +4 tokens (135.7 ms) cpos=96
950
+ [margins] 9.22 8.95 9.70 7.59 8.25 2.09 K=7
951
+ verify 21: accepted 6/6 (K=7) + correction -> +7 tokens (185.3 ms) cpos=103
952
+ [margins] 9.82 5.03 2.41 1.06 9.60 5.58 K=4
953
+ verify 22: accepted 3/3 (K=4) + correction -> +4 tokens (137.6 ms) cpos=107
954
+ [margins] 10.76 3.48 7.64 1.77 1.88 0.74 K=6
955
+ verify 23: accepted 3/5 (K=6) + correction -> +4 tokens (176.2 ms) cpos=111
956
+ [margins] 10.57 7.65 7.78 5.88 7.41 0.79 K=6
957
+ verify 24: accepted 5/5 (K=6) + correction -> +6 tokens (175.5 ms) cpos=117
958
+ [margins] 12.07 10.61 5.07 6.08 4.13 4.14 K=7
959
+ verify 25: accepted 6/6 (K=7) + correction -> +7 tokens (196.1 ms) cpos=124
960
+ [margins] 15.80 6.84 7.56 2.41 6.95 4.79 K=7
961
+ verify 26: accepted 6/6 (K=7) + correction -> +7 tokens (229.1 ms) cpos=131
962
+ [margins] 0.78 2.02 3.44 3.08 1.16 1.96 K=5
963
+ verify 27: accepted 0/4 (K=5) + correction -> +1 tokens (157.7 ms) cpos=132
964
+ [margins] 5.49 2.82 4.37 1.35 0.50 1.36 K=4
965
+ verify 28: accepted 3/3 (K=4) + correction -> +4 tokens (137.9 ms) cpos=136
966
+ [margins] 1.26 5.07 0.67 1.61 0.19 0.63 K=3
967
+ verify 29: accepted 2/2 (K=3) + correction -> +3 tokens (118.2 ms) cpos=139
968
+ [margins] 6.88 0.53 6.99 5.45 5.86 1.10 K=2
969
+ verify 30: accepted 1/1 (K=2) + correction -> +2 tokens (113.2 ms) cpos=141
970
+ [margins] 10.03 9.46 6.47 1.16 0.29 2.82 K=4
971
+ verify 31: accepted 3/3 (K=4) + correction -> +4 tokens (140.3 ms) cpos=145
972
+ [margins] 2.59 4.25 5.20 0.99 1.57 1.15 K=4
973
+ verify 32: accepted 0/3 (K=4) + correction -> +1 tokens (142.1 ms) cpos=146
974
+ [margins] 9.56 9.31 10.92 11.14 2.22 2.23 K=7
975
+ verify 33: accepted 4/6 (K=7) + correction -> +5 tokens (197.7 ms) cpos=151
976
+ [margins] 1.50 0.04 0.47 0.46 2.29 3.78 K=2
977
+ verify 34: accepted 1/1 (K=2) + correction -> +2 tokens (117.9 ms) cpos=153
978
+ [margins] 0.95 2.32 0.38 3.38 2.64 4.57 K=3
979
+ verify 35: accepted 0/2 (K=3) + correction -> +1 tokens (117.1 ms) cpos=154
980
+ [margins] 5.92 0.90 1.58 1.58 4.25 3.17 K=2
981
+ verify 36: accepted 1/1 (K=2) + correction -> +2 tokens (117.4 ms) cpos=156
982
+ [margins] 5.23 0.11 1.48 2.58 2.68 4.13 K=2
983
+ verify 37: accepted 1/1 (K=2) + correction -> +2 tokens (107.8 ms) cpos=158
984
+ [margins] 4.02 0.62 4.00 4.18 1.24 1.19 K=2
985
+ verify 38: accepted 1/1 (K=2) + correction -> +2 tokens (115.7 ms) cpos=160
986
+ [margins] 4.29 1.00 6.20 3.69 0.15 4.69 K=2
987
+ verify 39: accepted 1/1 (K=2) + correction -> +2 tokens (109.6 ms) cpos=162
988
+ [margins] 4.35 1.17 6.69 4.37 0.42 4.86 K=2
989
+ verify 40: accepted 1/1 (K=2) + correction -> +2 tokens (115.6 ms) cpos=164
990
+ [margins] 4.65 0.49 7.80 5.25 0.81 5.50 K=2
991
+ verify 41: accepted 1/1 (K=2) + correction -> +2 tokens (109.0 ms) cpos=166
992
+ [margins] 4.30 0.47 7.78 5.38 0.57 5.41 K=2
993
+ verify 42: accepted 1/1 (K=2) + correction -> +2 tokens (114.4 ms) cpos=168
994
+ [margins] 3.80 0.53 7.66 4.80 0.42 5.41 K=2
995
+ verify 43: accepted 1/1 (K=2) + correction -> +2 tokens (108.6 ms) cpos=170
996
+ [margins] 4.76 0.06 8.55 5.38 0.52 6.11 K=2
997
+ verify 44: accepted 1/1 (K=2) + correction -> +2 tokens (114.9 ms) cpos=172
998
+ [margins] 4.87 0.21 8.86 4.97 0.34 6.36 K=2
999
+ verify 45: accepted 1/1 (K=2) + correction -> +2 tokens (108.4 ms) cpos=174
1000
+ [margins] 3.83 0.12 7.60 4.81 0.04 6.51 K=2
1001
+ verify 46: accepted 1/1 (K=2) + correction -> +2 tokens (115.3 ms) cpos=176
1002
+ [margins] 4.25 0.37 7.86 5.62 0.01 6.32 K=2
1003
+ verify 47: accepted 1/1 (K=2) + correction -> +2 tokens (108.7 ms) cpos=178
1004
+ [margins] 4.08 0.14 8.10 5.12 0.18 6.80 K=2
1005
+ verify 48: accepted 1/1 (K=2) + correction -> +2 tokens (115.4 ms) cpos=180
1006
+ [margins] 4.28 0.44 8.33 5.27 0.13 6.21 K=2
1007
+ verify 49: accepted 1/1 (K=2) + correction -> +2 tokens (109.6 ms) cpos=182
1008
+ [margins] 5.18 0.90 8.46 5.46 0.50 6.27 K=2
1009
+ verify 50: accepted 1/1 (K=2) + correction -> +2 tokens (116.0 ms) cpos=184
1010
+ [margins] 5.19 0.11 7.82 5.34 0.56 5.95 K=2
1011
+ verify 51: accepted 1/1 (K=2) + correction -> +2 tokens (107.8 ms) cpos=186
1012
+ [margins] 4.48 0.02 7.90 5.39 0.19 5.83 K=2
1013
+ verify 52: accepted 1/1 (K=2) + correction -> +2 tokens (115.4 ms) cpos=188
1014
+ [margins] 4.82 0.07 7.77 5.41 0.80 5.93 K=2
1015
+ verify 53: accepted 1/1 (K=2) + correction -> +2 tokens (109.1 ms) cpos=190
1016
+ [margins] 4.47 0.07 7.66 5.39 0.74 5.96 K=2
1017
+ verify 54: accepted 1/1 (K=2) + correction -> +2 tokens (115.5 ms) cpos=192
1018
+ [margins] 4.25 0.08 6.30 4.86 0.42 5.76 K=2
1019
+ verify 55: accepted 1/1 (K=2) + correction -> +2 tokens (108.5 ms) cpos=194
1020
+ [margins] 4.78 0.59 8.23 5.06 1.69 6.04 K=2
1021
+ verify 56: accepted 1/1 (K=2) + correction -> +2 tokens (115.3 ms) cpos=196
1022
+ [margins] 3.87 0.36 7.79 4.66 0.99 6.03 K=2
1023
+ verify 57: accepted 1/1 (K=2) + correction -> +2 tokens (108.9 ms) cpos=198
1024
+ [margins] 3.90 0.01 7.74 4.98 0.77 6.11 K=2
1025
+ verify 58: accepted 1/1 (K=2) + correction -> +2 tokens (116.7 ms) cpos=200
1026
+ [margins] 4.47 0.16 7.68 5.00 1.05 6.03 K=2
1027
+ verify 59: accepted 1/1 (K=2) + correction -> +2 tokens (109.4 ms) cpos=202
1028
+ [margins] 4.90 0.58 6.43 5.27 1.25 5.98 K=2
1029
+ verify 60: accepted 1/1 (K=2) + correction -> +2 tokens (116.4 ms) cpos=204
1030
+ [margins] 4.63 0.59 6.22 5.28 1.33 5.89 K=2
1031
+ verify 61: accepted 1/1 (K=2) + correction -> +2 tokens (109.0 ms) cpos=206
1032
+ [margins] 4.84 0.36 7.12 5.19 1.09 6.68 K=2
1033
+ verify 62: accepted 1/1 (K=2) + correction -> +2 tokens (116.1 ms) cpos=208
1034
+ [margins] 4.18 0.16 6.89 5.30 1.49 6.32 K=2
1035
+ verify 63: accepted 1/1 (K=2) + correction -> +2 tokens (109.6 ms) cpos=210
1036
+ [margins] 5.04 1.25 7.09 5.09 1.65 6.36 K=2
1037
+ verify 64: accepted 1/1 (K=2) + correction -> +2 tokens (115.9 ms) cpos=212
1038
+ [margins] 4.63 0.15 6.96 5.20 1.49 6.52 K=2
1039
+ verify 65: accepted 1/1 (K=2) + correction -> +2 tokens (109.5 ms) cpos=214
1040
+ [margins] 4.74 0.44 6.85 4.97 0.43 6.22 K=2
1041
+ verify 66: accepted 1/1 (K=2) + correction -> +2 tokens (115.3 ms) cpos=216
1042
+ [margins] 4.30 0.66 6.45 4.79 1.72 6.06 K=2
1043
+ verify 67: accepted 1/1 (K=2) + correction -> +2 tokens (109.3 ms) cpos=218
1044
+ [margins] 4.49 1.03 6.55 5.08 1.37 5.90 K=2
1045
+ verify 68: accepted 1/1 (K=2) + correction -> +2 tokens (116.2 ms) cpos=220
1046
+ [margins] 4.34 0.79 6.70 4.95 1.10 5.91 K=2
1047
+ verify 69: accepted 1/1 (K=2) + correction -> +2 tokens (110.0 ms) cpos=222
1048
+ [margins] 4.93 0.96 7.08 4.92 1.39 6.45 K=2
1049
+ verify 70: accepted 1/1 (K=2) + correction -> +2 tokens (116.1 ms) cpos=224
1050
+ [margins] 4.44 0.08 6.80 4.86 0.71 6.33 K=2
1051
+ verify 71: accepted 1/1 (K=2) + correction -> +2 tokens (109.4 ms) cpos=226
1052
+ [margins] 4.96 1.05 7.04 5.49 1.24 6.36 K=2
1053
+ verify 72: accepted 1/1 (K=2) + correction -> +2 tokens (116.2 ms) cpos=228
1054
+ [margins] 5.03 0.88 6.65 5.35 1.26 6.18 K=2
1055
+ verify 73: accepted 1/1 (K=2) + correction -> +2 tokens (109.6 ms) cpos=230
1056
+ [margins] 4.27 1.23 6.87 5.04 0.90 6.34 K=2
1057
+ verify 74: accepted 0/1 (K=2) + correction -> +1 tokens (116.1 ms) cpos=231
1058
+ [margins] 8.21 3.94 1.17 7.84 5.44 0.29 K=3
1059
+ verify 75: accepted 0/2 (K=3) + correction -> +1 tokens (126.4 ms) cpos=232
1060
+ [margins] 5.90 2.37 6.95 3.01 0.73 6.55 K=5
1061
+ verify 76: accepted 1/4 (K=5) + correction -> +2 tokens (148.6 ms) cpos=234
1062
+ [margins] 2.78 2.86 1.32 1.17 0.31 6.14 K=3
1063
+ verify 77: accepted 1/2 (K=3) + correction -> +2 tokens (124.4 ms) cpos=236
1064
+ [margins] 3.75 0.55 6.33 0.62 5.79 1.33 K=2
1065
+ verify 78: accepted 0/1 (K=2) + correction -> +1 tokens (106.8 ms) cpos=237
1066
+
1067
+ [spec] generated 200 tokens over 78 verifies: mean tokens/verify = 2.56 (block=6, max 7)
1068
+ [spec] tokens: 343 12888 28 566 344 260 19075 3417 9508 5718 270 2753 294 270 6616 366 1512 1527 27819 13459 16 5715 270 12919 6329 1512 1527 940 223 19 27819 13459 16 455 3452 2753 344 979 7490 201
1069
+ [spec] SPEC-DECODE: 48.8 ms/tok = 20.48 tok/s (vs base M=1 70.7 ms/tok = 14.14 tok/s -> 1.45x)
1070
+ [decode] PREFILL: 12 positions in 257.5 ms = 46.6 tok/s (21.459 ms/tok)
1071
+ [spec] decoding 200 tokens (block=6, draft passes=1, adaptK=1.50, prompt=8 s=13)...
1072
+ [margins] 2.37 2.62 0.72 7.22 4.91 0.90 K=3
1073
+ verify 1: accepted 1/2 (K=3) + correction -> +2 tokens (114.2 ms) cpos=14
1074
+ [margins] 0.21 4.47 4.04 1.13 5.17 4.31 K=4
1075
+ verify 2: accepted 0/3 (K=4) + correction -> +1 tokens (133.1 ms) cpos=15
1076
+ [margins] 4.48 2.37 3.76 0.80 1.83 1.13 K=4
1077
+ verify 3: accepted 0/3 (K=4) + correction -> +1 tokens (124.4 ms) cpos=16
1078
+ [margins] 2.04 1.65 0.28 0.15 0.93 0.78 K=3
1079
+ verify 4: accepted 0/2 (K=3) + correction -> +1 tokens (106.3 ms) cpos=17
1080
+ [margins] 2.06 4.91 2.87 1.15 0.86 0.26 K=4
1081
+ verify 5: accepted 0/3 (K=4) + correction -> +1 tokens (124.6 ms) cpos=18
1082
+ [margins] 1.07 0.45 2.01 1.31 0.01 3.15 K=2
1083
+ verify 6: accepted 0/1 (K=2) + correction -> +1 tokens (104.7 ms) cpos=19
1084
+ [margins] 0.30 2.69 1.04 0.39 0.15 3.81 K=3
1085
+ verify 7: accepted 0/2 (K=3) + correction -> +1 tokens (108.3 ms) cpos=20
1086
+ [margins] 0.61 4.17 0.18 0.46 1.48 0.83 K=3
1087
+ verify 8: accepted 0/2 (K=3) + correction -> +1 tokens (107.4 ms) cpos=21
1088
+ [margins] 0.26 2.19 0.65 0.91 3.28 0.62 K=3
1089
+ verify 9: accepted 0/2 (K=3) + correction -> +1 tokens (108.6 ms) cpos=22
1090
+ [margins] 0.67 4.95 0.32 3.67 0.69 0.35 K=3
1091
+ verify 10: accepted 0/2 (K=3) + correction -> +1 tokens (114.3 ms) cpos=23
1092
+ [margins] 0.36 1.47 0.06 0.09 0.09 3.20 K=2
1093
+ verify 11: accepted 0/1 (K=2) + correction -> +1 tokens (99.6 ms) cpos=24
1094
+ [margins] 0.27 5.01 0.30 4.02 0.84 0.07 K=3
1095
+ verify 12: accepted 0/2 (K=3) + correction -> +1 tokens (108.2 ms) cpos=25
1096
+ [margins] 0.67 1.37 0.26 0.07 0.40 4.02 K=2
1097
+ verify 13: accepted 1/1 (K=2) + correction -> +2 tokens (101.6 ms) cpos=27
1098
+ [margins] 1.65 0.83 0.94 0.04 1.07 0.65 K=2
1099
+ verify 14: accepted 1/1 (K=2) + correction -> +2 tokens (108.8 ms) cpos=29
1100
+ [margins] 2.93 1.68 2.66 0.19 1.03 1.54 K=4
1101
+ verify 15: accepted 3/3 (K=4) + correction -> +4 tokens (123.7 ms) cpos=33
1102
+ [margins] 4.27 1.42 3.21 0.16 1.74 0.92 K=2
1103
+ verify 16: accepted 1/1 (K=2) + correction -> +2 tokens (101.6 ms) cpos=35
1104
+ [margins] 5.13 1.09 5.21 0.01 3.00 0.83 K=2
1105
+ verify 17: accepted 1/1 (K=2) + correction -> +2 tokens (109.0 ms) cpos=37
1106
+ [margins] 4.76 1.18 4.62 0.24 3.05 0.65 K=2
1107
+ verify 18: accepted 1/1 (K=2) + correction -> +2 tokens (101.8 ms) cpos=39
1108
+ [margins] 5.59 1.02 5.06 0.20 2.85 1.05 K=2
1109
+ verify 19: accepted 1/1 (K=2) + correction -> +2 tokens (109.2 ms) cpos=41
1110
+ [margins] 5.42 1.44 4.93 0.13 2.80 0.53 K=2
1111
+ verify 20: accepted 1/1 (K=2) + correction -> +2 tokens (102.7 ms) cpos=43
1112
+ [margins] 5.76 0.59 5.56 0.37 5.10 1.36 K=2
1113
+ verify 21: accepted 1/1 (K=2) + correction -> +2 tokens (109.8 ms) cpos=45
1114
+ [margins] 5.79 1.04 4.38 1.18 3.50 0.22 K=2
1115
+ verify 22: accepted 0/1 (K=2) + correction -> +1 tokens (100.1 ms) cpos=46
1116
+ [margins] 4.03 2.27 3.67 0.05 0.80 0.34 K=4
1117
+ verify 23: accepted 3/3 (K=4) + correction -> +4 tokens (132.3 ms) cpos=50
1118
+ [margins] 3.14 3.52 0.02 3.30 4.94 1.14 K=3
1119
+ verify 24: accepted 2/2 (K=3) + correction -> +3 tokens (120.4 ms) cpos=53
1120
+ [margins] 8.45 8.51 0.48 5.44 0.83 1.96 K=3
1121
+ verify 25: accepted 2/2 (K=3) + correction -> +3 tokens (118.5 ms) cpos=56
1122
+ [margins] 2.68 0.59 0.57 1.76 1.42 2.81 K=2
1123
+ verify 26: accepted 1/1 (K=2) + correction -> +2 tokens (103.3 ms) cpos=58
1124
+ [margins] 0.71 4.85 0.92 0.80 1.82 0.86 K=3
1125
+ verify 27: accepted 0/2 (K=3) + correction -> +1 tokens (116.6 ms) cpos=59
1126
+ [margins] 6.25 0.75 1.57 5.63 0.78 0.05 K=2
1127
+ verify 28: accepted 1/1 (K=2) + correction -> +2 tokens (107.3 ms) cpos=61
1128
+ [margins] 6.97 7.10 6.45 4.06 8.76 2.17 K=7
1129
+ verify 29: accepted 6/6 (K=7) + correction -> +7 tokens (185.6 ms) cpos=68
1130
+ [margins] 9.00 2.73 3.45 0.08 0.04 3.89 K=4
1131
+ verify 30: accepted 2/3 (K=4) + correction -> +3 tokens (134.3 ms) cpos=71
1132
+ [margins] 1.97 4.36 2.19 6.41 4.98 4.58 K=7
1133
+ verify 31: accepted 1/6 (K=7) + correction -> +2 tokens (183.6 ms) cpos=73
1134
+ [margins] 6.49 8.15 6.38 2.99 7.87 0.08 K=6
1135
+ verify 32: accepted 5/5 (K=6) + correction -> +6 tokens (163.0 ms) cpos=79
1136
+ [margins] 7.18 9.48 0.72 0.13 3.99 1.21 K=3
1137
+ verify 33: accepted 2/2 (K=3) + correction -> +3 tokens (122.6 ms) cpos=82
1138
+ [margins] 4.27 7.45 1.78 5.23 1.08 0.20 K=5
1139
+ verify 34: accepted 4/4 (K=5) + correction -> +5 tokens (146.1 ms) cpos=87
1140
+ [margins] 9.14 4.83 4.54 0.70 5.30 3.04 K=4
1141
+ verify 35: accepted 3/3 (K=4) + correction -> +4 tokens (135.5 ms) cpos=91
1142
+ [margins] 8.63 11.72 1.25 3.28 7.15 5.15 K=3
1143
+ verify 36: accepted 2/2 (K=3) + correction -> +3 tokens (123.3 ms) cpos=94
1144
+ [margins] 6.97 9.24 7.18 5.90 0.66 2.60 K=5
1145
+ verify 37: accepted 4/4 (K=5) + correction -> +5 tokens (148.1 ms) cpos=99
1146
+ [margins] 9.35 4.02 4.65 1.99 3.20 1.46 K=6
1147
+ verify 38: accepted 5/5 (K=6) + correction -> +6 tokens (174.6 ms) cpos=105
1148
+ [margins] 2.82 5.51 11.13 4.37 7.60 4.00 K=7
1149
+ verify 39: accepted 6/6 (K=7) + correction -> +7 tokens (193.0 ms) cpos=112
1150
+ [margins] 3.89 5.49 3.52 3.04 5.38 6.12 K=7
1151
+ verify 40: accepted 2/6 (K=7) + correction -> +3 tokens (184.9 ms) cpos=115
1152
+ [margins] 8.38 12.63 2.84 7.38 10.65 6.45 K=7
1153
+ verify 41: accepted 6/6 (K=7) + correction -> +7 tokens (194.3 ms) cpos=122
1154
+ [margins] 9.69 5.91 0.73 0.45 1.57 0.92 K=3
1155
+ verify 42: accepted 2/2 (K=3) + correction -> +3 tokens (126.2 ms) cpos=125
1156
+ [margins] 10.89 8.72 8.61 9.10 3.18 6.17 K=7
1157
+ verify 43: accepted 6/6 (K=7) + correction -> +7 tokens (235.5 ms) cpos=132
1158
+ [margins] 5.47 10.49 9.16 9.30 3.04 8.59 K=7
1159
+ verify 44: accepted 6/6 (K=7) + correction -> +7 tokens (189.0 ms) cpos=139
1160
+ [margins] 5.13 8.21 3.04 6.92 6.61 0.74 K=6
1161
+ verify 45: accepted 5/5 (K=6) + correction -> +6 tokens (175.6 ms) cpos=145
1162
+ [margins] 12.76 11.05 7.81 2.87 4.58 2.82 K=7
1163
+ verify 46: accepted 5/6 (K=7) + correction -> +6 tokens (195.6 ms) cpos=151
1164
+ [margins] 8.33 11.97 2.49 7.76 9.14 4.94 K=7
1165
+ verify 47: accepted 6/6 (K=7) + correction -> +7 tokens (195.3 ms) cpos=158
1166
+ [margins] 8.96 7.41 1.38 6.00 4.19 6.75 K=3
1167
+ verify 48: accepted 2/2 (K=3) + correction -> +3 tokens (128.9 ms) cpos=161
1168
+ [margins] 11.24 7.29 11.64 11.51 1.56 5.20 K=7
1169
+ verify 49: accepted 6/6 (K=7) + correction -> +7 tokens (194.7 ms) cpos=168
1170
+ [margins] 5.65 11.32 8.70 9.29 5.92 9.41 K=7
1171
+ verify 50: accepted 6/6 (K=7) + correction -> +7 tokens (188.9 ms) cpos=175
1172
+ [margins] 7.99 10.90 2.86 8.56 7.70 6.74 K=7
1173
+ verify 51: accepted 6/6 (K=7) + correction -> +7 tokens (196.0 ms) cpos=182
1174
+ [margins] 12.84 5.80 1.10 4.32 5.25 7.11 K=3
1175
+ verify 52: accepted 2/2 (K=3) + correction -> +3 tokens (128.6 ms) cpos=185
1176
+ [margins] 11.64 8.01 12.00 11.12 3.19 6.65 K=7
1177
+ verify 53: accepted 6/6 (K=7) + correction -> +7 tokens (196.7 ms) cpos=192
1178
+ [margins] 5.20 11.42 8.74 9.07 2.82 9.44 K=7
1179
+ verify 54: accepted 6/6 (K=7) + correction -> +7 tokens (190.5 ms) cpos=199
1180
+ [margins] 9.23 12.90 3.54 6.06 9.72 5.64 K=7
1181
+ verify 55: accepted 6/6 (K=7) + correction -> +7 tokens (198.2 ms) cpos=206
1182
+ [margins] 10.53 7.60 1.17 6.33 4.01 4.23 K=3
1183
+ verify 56: accepted 2/2 (K=3) + correction -> +3 tokens (127.8 ms) cpos=209
1184
+ [margins] 12.46 6.55 9.64 9.71 2.68 7.72 K=7
1185
+ verify 57: accepted 6/6 (K=7) + correction -> +7 tokens (196.5 ms) cpos=216
1186
+
1187
+ [spec] generated 204 tokens over 57 verifies: mean tokens/verify = 3.58 (block=6, max 7)
1188
+ [spec] tokens: 1 201 1 201 1 201 1 201 1 201 1 201 1 201 1 201 1 201 1 201 1 201 1 201 1 201 1 201 1 201 1 201 3085 344 270 6102 294 8760 33 9361
1189
+ [spec] SPEC-DECODE: 39.4 ms/tok = 25.38 tok/s (vs base M=1 70.7 ms/tok = 14.14 tok/s -> 1.80x)
1190
+
1191
+ [blksweep] BLK passes adaptK->eff prompt | mean tok/verify | ms/tok | tok/s | vs base 14.14 tok/s
1192
+ [blksweep] 6 1 1.50 1.50 0 | 3.35 | 42.0 | 23.82 | 1.68x
1193
+ [blksweep] 6 1 1.50 1.50 1 | 4.64 | 33.8 | 29.59 | 2.09x
1194
+ [blksweep] 6 1 1.50 1.50 2 | 4.21 | 36.7 | 27.28 | 1.93x
1195
+ [blksweep] 6 1 1.50 1.50 3 | 2.15 | 56.7 | 17.62 | 1.25x
1196
+ [blksweep] 6 1 1.50 1.50 4 | 2.37 | 51.1 | 19.59 | 1.39x
1197
+ [blksweep] 6 1 1.50 1.50 5 | 4.35 | 37.2 | 26.89 | 1.90x
1198
+ [blksweep] 6 1 1.50 1.50 6 | 5.26 | 30.9 | 32.35 | 2.29x
1199
+ [blksweep] 6 1 1.50 1.50 7 | 2.56 | 48.8 | 20.48 | 1.45x
1200
+ [blksweep] 6 1 1.50 1.50 8 | 3.58 | 39.4 | 25.38 | 1.80x
1201
+ [decode] mem 107.9/122.8 GiB