patrickbdevaney commited on
Commit
9bc18dc
·
verified ·
1 Parent(s): 139bad5

archive arm s3recap-conf1.0: eval.log

Browse files
Files changed (1) hide show
  1. arms/s3recap-conf1.0/eval.log +1105 -0
arms/s3recap-conf1.0/eval.log ADDED
@@ -0,0 +1,1105 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [decode] loading /home/patrickd/models/DeepSeek-V4-Flash-0731-REAP ... s=6 NDEC=8 seqmax=297
2
+ [decode] loaded 100.40 GiB, 45821 tensors (weights in MANAGED device-preferred memory)
3
+ [decode] KV layout: fp32, 2048 B/row, 0.024 GiB for 297 rows x 43 layers
4
+ [decode] building 43 layer structs once (persistent)...
5
+ [decode] structs built. mem 106.8/122.8 GiB
6
+ [decode] prefill 5 positions...
7
+ [f32mk] tile 4x4 (DSV4_F32MK_TILE=unset)
8
+ [decode] PREFILL: 5 positions in 194.6 ms = 25.7 tok/s (38.914 ms/tok)
9
+ [decode] prefill done. caches populated. starting decode.
10
+ step 0 pos 5 -> token 11111 (66.9 ms warmup)
11
+ step 1 pos 6 -> token 16 (66.8 ms)
12
+ step 2 pos 7 -> token 455 (72.7 ms)
13
+ step 3 pos 8 -> token 6102 (66.6 ms)
14
+ step 4 pos 9 -> token 294 (67.2 ms)
15
+ step 5 pos 10 -> token 16603 (67.0 ms)
16
+ step 6 pos 11 -> token 344 (73.0 ms)
17
+ step 7 pos 12 -> token 29168 (67.0 ms)
18
+
19
+ [decode] first decoded token argmax = 11111 (expect 11111) -> GATE PASS
20
+ [decode] generated: 11111 16 455 6102 294 16603 344 29168
21
+ [decode] WARM decode: 68.6 ms/tok = 14.57 tok/s (M=1 steady state, 7-step avg)
22
+
23
+ [spec-verify] M=5 verify in ONE forward: 115.7 ms (= 23.1 ms/tok if all accepted vs 68.6 M=1 -> 2.96x)
24
+ [spec-verify] verify argmax: 11111 16 455 6102 294
25
+ [spec-verify] decode tokens: 11111 16 455 6102 294
26
+ [spec-verify] MATCH 5/5 -> PASS (M=K verify == K sequential decodes; diffs = MoE-atomic near-ties)
27
+
28
+ [spec] loading EXTERNAL DSpark head /home/patrickd/s5-capture/s3recap-conf1.0/head ...
29
+ [spec] head loaded 6.53 GiB, 2977 mtp tensors (SEPARATE store)
30
+ [spec] NSTAGE=3 head-experts=160 BLK sweep: 5 5 5 5 5 5 5 5 5
31
+ [spec] head built. mem 112.7/122.8 GiB
32
+ [decode] PREFILL: 5 positions in 150.1 ms = 33.3 tok/s (30.029 ms/tok)
33
+ [spec] decoding 200 tokens (block=5, draft passes=1, adaptK=1.50, prompt=0 s=6)...
34
+ [arena] first draft: draft footprint 210.72 MB, global hwm 210.72 MB / cap 512.00 MB (41.2%), draft path = ARENA
35
+ [margins] 1.47 0.95 0.52 0.62 0.05 K=2
36
+ verify 1: accepted 1/1 (K=2) + correction -> +2 tokens (100.7 ms) cpos=7
37
+ [margins] 0.57 0.13 0.62 0.46 0.19 K=2
38
+ verify 2: accepted 0/1 (K=2) + correction -> +1 tokens (104.7 ms) cpos=8
39
+ [margins] 0.53 0.67 2.15 0.75 0.18 K=2
40
+ verify 3: accepted 1/1 (K=2) + correction -> +2 tokens (97.7 ms) cpos=10
41
+ [margins] 2.42 0.72 4.06 0.13 1.63 K=2
42
+ verify 4: accepted 0/1 (K=2) + correction -> +1 tokens (102.6 ms) cpos=11
43
+ [margins] 6.74 0.52 0.77 2.98 2.42 K=2
44
+ verify 5: accepted 1/1 (K=2) + correction -> +2 tokens (103.0 ms) cpos=13
45
+ [margins] 4.46 1.86 5.04 2.56 0.25 K=5
46
+ verify 6: accepted 4/4 (K=5) + correction -> +5 tokens (141.3 ms) cpos=18
47
+ [margins] 4.67 2.52 2.75 0.83 2.68 K=4
48
+ verify 7: accepted 1/3 (K=4) + correction -> +2 tokens (126.8 ms) cpos=20
49
+ [margins] 2.94 3.16 4.72 2.56 0.68 K=5
50
+ verify 8: accepted 4/4 (K=5) + correction -> +5 tokens (140.3 ms) cpos=25
51
+ [margins] 7.02 2.26 3.35 0.51 0.17 K=4
52
+ verify 9: accepted 1/3 (K=4) + correction -> +2 tokens (127.4 ms) cpos=27
53
+ [margins] 4.10 3.33 4.82 2.91 0.55 K=5
54
+ verify 10: accepted 4/4 (K=5) + correction -> +5 tokens (147.2 ms) cpos=32
55
+ [margins] 7.48 0.10 2.52 1.29 2.62 K=2
56
+ verify 11: accepted 1/1 (K=2) + correction -> +2 tokens (97.1 ms) cpos=34
57
+ [margins] 2.52 4.47 5.57 2.16 0.23 K=5
58
+ verify 12: accepted 4/4 (K=5) + correction -> +5 tokens (141.8 ms) cpos=39
59
+ [margins] 4.25 0.16 5.00 2.00 3.59 K=2
60
+ verify 13: accepted 1/1 (K=2) + correction -> +2 tokens (101.3 ms) cpos=41
61
+ [margins] 4.12 4.43 4.64 1.08 2.23 K=4
62
+ verify 14: accepted 3/3 (K=4) + correction -> +4 tokens (127.5 ms) cpos=45
63
+ [margins] 2.87 0.50 7.12 1.50 4.52 K=2
64
+ verify 15: accepted 1/1 (K=2) + correction -> +2 tokens (97.9 ms) cpos=47
65
+ [margins] 0.87 3.79 0.69 4.72 2.45 K=3
66
+ verify 16: accepted 2/2 (K=3) + correction -> +3 tokens (113.8 ms) cpos=50
67
+ [margins] 4.90 3.36 3.76 0.92 2.55 K=4
68
+ verify 17: accepted 3/3 (K=4) + correction -> +4 tokens (132.6 ms) cpos=54
69
+ [margins] 2.75 1.30 6.55 0.93 0.81 K=2
70
+ verify 18: accepted 1/1 (K=2) + correction -> +2 tokens (101.7 ms) cpos=56
71
+ [margins] 8.84 0.20 4.14 2.81 2.33 K=2
72
+ verify 19: accepted 1/1 (K=2) + correction -> +2 tokens (98.0 ms) cpos=58
73
+ [margins] 2.43 5.35 5.58 4.39 1.84 K=6
74
+ verify 20: accepted 0/5 (K=6) + correction -> +1 tokens (164.8 ms) cpos=59
75
+ [margins] 6.14 7.50 5.64 0.94 5.90 K=4
76
+ verify 21: accepted 3/3 (K=4) + correction -> +4 tokens (129.9 ms) cpos=63
77
+ [margins] 0.46 0.36 7.31 1.38 2.52 K=2
78
+ verify 22: accepted 0/1 (K=2) + correction -> +1 tokens (102.7 ms) cpos=64
79
+ [margins] 7.07 1.14 4.25 1.21 4.41 K=2
80
+ verify 23: accepted 1/1 (K=2) + correction -> +2 tokens (99.2 ms) cpos=66
81
+ [margins] 4.86 5.29 4.32 3.20 1.17 K=5
82
+ verify 24: accepted 4/4 (K=5) + correction -> +5 tokens (141.5 ms) cpos=71
83
+ [margins] 6.36 1.97 3.24 0.65 2.79 K=4
84
+ verify 25: accepted 3/3 (K=4) + correction -> +4 tokens (130.7 ms) cpos=75
85
+ [margins] 5.80 6.63 2.72 0.49 2.72 K=4
86
+ verify 26: accepted 2/3 (K=4) + correction -> +3 tokens (129.5 ms) cpos=78
87
+ [margins] 5.26 0.23 2.48 1.05 3.19 K=2
88
+ verify 27: accepted 1/1 (K=2) + correction -> +2 tokens (103.5 ms) cpos=80
89
+ [margins] 5.70 6.20 5.03 4.38 1.21 K=5
90
+ verify 28: accepted 4/4 (K=5) + correction -> +5 tokens (144.9 ms) cpos=85
91
+ [margins] 5.53 1.24 3.87 1.95 2.92 K=2
92
+ verify 29: accepted 1/1 (K=2) + correction -> +2 tokens (100.2 ms) cpos=87
93
+ [margins] 5.44 5.19 5.03 2.64 0.65 K=5
94
+ verify 30: accepted 4/4 (K=5) + correction -> +5 tokens (150.5 ms) cpos=92
95
+ [margins] 6.86 1.09 2.83 0.30 1.93 K=2
96
+ verify 31: accepted 1/1 (K=2) + correction -> +2 tokens (101.4 ms) cpos=94
97
+ [margins] 5.17 6.46 5.21 5.76 0.67 K=5
98
+ verify 32: accepted 4/4 (K=5) + correction -> +5 tokens (147.0 ms) cpos=99
99
+ [margins] 5.19 1.40 1.13 0.58 0.91 K=2
100
+ verify 33: accepted 1/1 (K=2) + correction -> +2 tokens (107.3 ms) cpos=101
101
+ [margins] 7.02 4.11 4.37 2.00 3.53 K=6
102
+ verify 34: accepted 4/5 (K=6) + correction -> +5 tokens (161.8 ms) cpos=106
103
+ [margins] 6.12 1.17 2.82 2.04 2.58 K=2
104
+ verify 35: accepted 1/1 (K=2) + correction -> +2 tokens (106.3 ms) cpos=108
105
+ [margins] 7.28 5.51 6.63 3.27 0.77 K=5
106
+ verify 36: accepted 4/4 (K=5) + correction -> +5 tokens (144.6 ms) cpos=113
107
+ [margins] 6.77 1.35 2.63 1.17 3.36 K=2
108
+ verify 37: accepted 1/1 (K=2) + correction -> +2 tokens (101.7 ms) cpos=115
109
+ [margins] 4.15 3.02 7.74 3.32 0.44 K=5
110
+ verify 38: accepted 4/4 (K=5) + correction -> +5 tokens (150.5 ms) cpos=120
111
+ [margins] 6.82 0.76 1.23 1.03 0.54 K=2
112
+ verify 39: accepted 1/1 (K=2) + correction -> +2 tokens (102.6 ms) cpos=122
113
+ [margins] 5.52 4.28 4.81 4.73 1.83 K=6
114
+ verify 40: accepted 4/5 (K=6) + correction -> +5 tokens (184.2 ms) cpos=127
115
+ [margins] 6.72 1.42 0.76 1.72 2.71 K=2
116
+ verify 41: accepted 1/1 (K=2) + correction -> +2 tokens (120.3 ms) cpos=129
117
+ [margins] 8.22 4.18 7.00 1.69 2.13 K=6
118
+ verify 42: accepted 4/5 (K=6) + correction -> +5 tokens (161.5 ms) cpos=134
119
+ [margins] 5.38 1.44 0.07 2.37 0.06 K=2
120
+ verify 43: accepted 1/1 (K=2) + correction -> +2 tokens (108.3 ms) cpos=136
121
+ [margins] 7.92 5.49 6.24 2.92 0.99 K=5
122
+ verify 44: accepted 4/4 (K=5) + correction -> +5 tokens (145.8 ms) cpos=141
123
+ [margins] 5.74 4.75 4.48 2.38 0.42 K=5
124
+ verify 45: accepted 3/4 (K=5) + correction -> +4 tokens (142.9 ms) cpos=145
125
+ [margins] 5.11 5.49 6.65 2.11 0.19 K=5
126
+ verify 46: accepted 4/4 (K=5) + correction -> +5 tokens (144.6 ms) cpos=150
127
+ [margins] 6.12 0.87 3.19 1.04 4.85 K=2
128
+ verify 47: accepted 1/1 (K=2) + correction -> +2 tokens (106.8 ms) cpos=152
129
+ [margins] 16.85 2.26 7.90 6.93 9.32 K=6
130
+ verify 48: accepted 1/5 (K=6) + correction -> +2 tokens (162.6 ms) cpos=154
131
+ [margins] 4.67 9.58 5.65 5.67 3.12 K=6
132
+ verify 49: accepted 4/5 (K=6) + correction -> +5 tokens (167.4 ms) cpos=159
133
+ [margins] 3.89 0.43 3.22 2.06 1.95 K=2
134
+ verify 50: accepted 1/1 (K=2) + correction -> +2 tokens (105.8 ms) cpos=161
135
+ [margins] 8.80 6.55 8.25 6.20 3.80 K=6
136
+ verify 51: accepted 5/5 (K=6) + correction -> +6 tokens (164.3 ms) cpos=167
137
+ [margins] 5.39 1.13 3.65 2.62 0.45 K=2
138
+ verify 52: accepted 1/1 (K=2) + correction -> +2 tokens (105.2 ms) cpos=169
139
+ [margins] 2.83 5.73 2.60 3.78 1.87 K=6
140
+ verify 53: accepted 4/5 (K=6) + correction -> +5 tokens (161.4 ms) cpos=174
141
+ [margins] 3.37 1.79 4.23 1.81 3.44 K=6
142
+ verify 54: accepted 1/5 (K=6) + correction -> +2 tokens (169.3 ms) cpos=176
143
+ [margins] 4.20 4.83 6.29 3.31 0.32 K=5
144
+ verify 55: accepted 4/4 (K=5) + correction -> +5 tokens (145.1 ms) cpos=181
145
+ [margins] 5.92 1.45 2.38 3.10 4.02 K=2
146
+ verify 56: accepted 1/1 (K=2) + correction -> +2 tokens (103.2 ms) cpos=183
147
+ [margins] 4.30 5.57 7.92 3.28 1.12 K=5
148
+ verify 57: accepted 4/4 (K=5) + correction -> +5 tokens (151.3 ms) cpos=188
149
+ [margins] 2.80 0.59 1.67 0.90 3.63 K=2
150
+ verify 58: accepted 1/1 (K=2) + correction -> +2 tokens (103.4 ms) cpos=190
151
+ [margins] 9.57 2.20 1.58 1.82 3.21 K=6
152
+ verify 59: accepted 1/5 (K=6) + correction -> +2 tokens (169.3 ms) cpos=192
153
+ [margins] 5.06 3.22 6.36 3.78 8.67 K=6
154
+ verify 60: accepted 5/5 (K=6) + correction -> +6 tokens (163.0 ms) cpos=198
155
+ [margins] 1.71 5.56 0.84 1.93 3.82 K=3
156
+ verify 61: accepted 2/2 (K=3) + correction -> +3 tokens (120.2 ms) cpos=201
157
+ [margins] 5.64 9.06 7.78 7.19 2.60 K=6
158
+ verify 62: accepted 5/5 (K=6) + correction -> +6 tokens (160.8 ms) cpos=207
159
+
160
+ [spec] generated 202 tokens over 62 verifies: mean tokens/verify = 3.26 (block=5, max 6)
161
+ [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
162
+ [spec] LOSSLESS GATE: first 8 tokens match base AR -> PASS
163
+ [spec] SPEC-DECODE: 39.5 ms/tok = 25.35 tok/s (vs base M=1 68.6 ms/tok = 14.57 tok/s -> 1.74x)
164
+ [decode] PREFILL: 34 positions in 603.7 ms = 56.3 tok/s (17.756 ms/tok)
165
+ [spec] decoding 200 tokens (block=5, draft passes=1, adaptK=1.50, prompt=1 s=35)...
166
+ [margins] 1.59 0.48 1.18 1.52 0.35 K=2
167
+ verify 1: accepted 0/1 (K=2) + correction -> +1 tokens (104.4 ms) cpos=35
168
+ [margins] 2.65 2.17 1.61 2.02 4.56 K=6
169
+ verify 2: accepted 0/5 (K=6) + correction -> +1 tokens (159.0 ms) cpos=36
170
+ [margins] 1.24 1.24 0.61 3.51 0.99 K=2
171
+ verify 3: accepted 0/1 (K=2) + correction -> +1 tokens (96.4 ms) cpos=37
172
+ [margins] 1.29 2.86 1.95 1.44 0.68 K=4
173
+ verify 4: accepted 2/3 (K=4) + correction -> +3 tokens (123.7 ms) cpos=40
174
+ [margins] 1.13 0.44 0.21 0.56 1.08 K=2
175
+ verify 5: accepted 0/1 (K=2) + correction -> +1 tokens (97.4 ms) cpos=41
176
+ [margins] 1.43 4.00 0.03 0.04 0.35 K=3
177
+ verify 6: accepted 2/2 (K=3) + correction -> +3 tokens (112.6 ms) cpos=44
178
+ [margins] 0.38 5.11 10.01 2.18 0.88 K=5
179
+ verify 7: accepted 4/4 (K=5) + correction -> +5 tokens (138.8 ms) cpos=49
180
+ [margins] 5.80 2.89 0.82 1.82 1.12 K=3
181
+ verify 8: accepted 2/2 (K=3) + correction -> +3 tokens (111.0 ms) cpos=52
182
+ [margins] 4.21 5.44 7.06 2.04 1.01 K=5
183
+ verify 9: accepted 3/4 (K=5) + correction -> +4 tokens (141.5 ms) cpos=56
184
+ [margins] 3.16 3.04 1.01 0.31 2.97 K=3
185
+ verify 10: accepted 2/2 (K=3) + correction -> +3 tokens (109.3 ms) cpos=59
186
+ [margins] 0.68 0.73 2.77 0.03 4.87 K=2
187
+ verify 11: accepted 0/1 (K=2) + correction -> +1 tokens (102.5 ms) cpos=60
188
+ [margins] 1.46 1.71 5.32 3.85 1.62 K=6
189
+ verify 12: accepted 0/5 (K=6) + correction -> +1 tokens (150.3 ms) cpos=61
190
+ [margins] 0.78 2.36 2.07 1.66 0.58 K=5
191
+ verify 13: accepted 1/4 (K=5) + correction -> +2 tokens (134.2 ms) cpos=63
192
+ [margins] 2.09 3.51 1.26 4.56 0.27 K=3
193
+ verify 14: accepted 0/2 (K=3) + correction -> +1 tokens (115.7 ms) cpos=64
194
+ [margins] 3.41 0.07 0.79 5.95 0.54 K=2
195
+ verify 15: accepted 1/1 (K=2) + correction -> +2 tokens (98.2 ms) cpos=66
196
+ [margins] 1.06 0.40 3.37 4.80 0.30 K=2
197
+ verify 16: accepted 1/1 (K=2) + correction -> +2 tokens (105.3 ms) cpos=68
198
+ [margins] 0.05 5.82 0.14 2.35 1.47 K=3
199
+ verify 17: accepted 2/2 (K=3) + correction -> +3 tokens (111.9 ms) cpos=71
200
+ [margins] 2.33 1.44 1.39 4.93 0.30 K=2
201
+ verify 18: accepted 1/1 (K=2) + correction -> +2 tokens (105.5 ms) cpos=73
202
+ [margins] 1.27 0.27 0.29 5.23 0.94 K=2
203
+ verify 19: accepted 0/1 (K=2) + correction -> +1 tokens (100.2 ms) cpos=74
204
+ [margins] 7.12 1.68 3.62 2.19 2.73 K=6
205
+ verify 20: accepted 5/5 (K=6) + correction -> +6 tokens (161.5 ms) cpos=80
206
+ [margins] 4.88 4.98 3.98 2.09 7.74 K=6
207
+ verify 21: accepted 5/5 (K=6) + correction -> +6 tokens (159.6 ms) cpos=86
208
+ [margins] 7.79 2.73 2.54 6.35 1.13 K=5
209
+ verify 22: accepted 4/4 (K=5) + correction -> +5 tokens (144.0 ms) cpos=91
210
+ [margins] 7.09 2.31 3.04 6.96 1.21 K=5
211
+ verify 23: accepted 4/4 (K=5) + correction -> +5 tokens (147.9 ms) cpos=96
212
+ [margins] 9.53 4.55 3.51 8.86 3.82 K=6
213
+ verify 24: accepted 5/5 (K=6) + correction -> +6 tokens (158.7 ms) cpos=102
214
+ [margins] 10.30 3.61 6.80 6.55 5.60 K=6
215
+ verify 25: accepted 5/5 (K=6) + correction -> +6 tokens (166.6 ms) cpos=108
216
+ [margins] 2.94 0.20 0.69 0.52 2.12 K=2
217
+ verify 26: accepted 0/1 (K=2) + correction -> +1 tokens (102.4 ms) cpos=109
218
+ [margins] 7.92 5.86 4.55 6.39 5.68 K=6
219
+ verify 27: accepted 5/5 (K=6) + correction -> +6 tokens (160.2 ms) cpos=115
220
+ [margins] 9.32 8.53 8.10 2.64 7.84 K=6
221
+ verify 28: accepted 5/5 (K=6) + correction -> +6 tokens (168.1 ms) cpos=121
222
+ [margins] 8.24 4.04 7.47 6.80 0.10 K=5
223
+ verify 29: accepted 4/4 (K=5) + correction -> +5 tokens (148.0 ms) cpos=126
224
+ [margins] 9.45 5.91 6.34 1.75 4.61 K=6
225
+ verify 30: accepted 3/5 (K=6) + correction -> +4 tokens (184.6 ms) cpos=130
226
+ [margins] 8.59 7.45 6.34 2.82 10.02 K=6
227
+ verify 31: accepted 5/5 (K=6) + correction -> +6 tokens (167.2 ms) cpos=136
228
+ [margins] 7.57 4.48 5.77 8.67 3.23 K=6
229
+ verify 32: accepted 5/5 (K=6) + correction -> +6 tokens (163.1 ms) cpos=142
230
+ [margins] 13.77 7.00 9.49 11.53 8.32 K=6
231
+ verify 33: accepted 5/5 (K=6) + correction -> +6 tokens (168.7 ms) cpos=148
232
+ [margins] 3.30 1.13 8.10 4.56 3.59 K=2
233
+ verify 34: accepted 0/1 (K=2) + correction -> +1 tokens (104.0 ms) cpos=149
234
+ [margins] 7.32 9.81 7.85 8.27 5.18 K=6
235
+ verify 35: accepted 5/5 (K=6) + correction -> +6 tokens (161.1 ms) cpos=155
236
+ [margins] 10.89 9.00 7.57 1.94 8.16 K=6
237
+ verify 36: accepted 5/5 (K=6) + correction -> +6 tokens (169.1 ms) cpos=161
238
+ [margins] 10.87 5.60 3.64 9.12 7.38 K=6
239
+ verify 37: accepted 5/5 (K=6) + correction -> +6 tokens (162.3 ms) cpos=167
240
+ [margins] 14.03 7.11 9.17 10.62 10.26 K=6
241
+ verify 38: accepted 5/5 (K=6) + correction -> +6 tokens (169.5 ms) cpos=173
242
+ [margins] 2.54 7.74 10.88 6.19 6.10 K=6
243
+ verify 39: accepted 5/5 (K=6) + correction -> +6 tokens (160.9 ms) cpos=179
244
+ [margins] 9.04 10.55 6.98 9.57 6.02 K=6
245
+ verify 40: accepted 5/5 (K=6) + correction -> +6 tokens (166.8 ms) cpos=185
246
+ [margins] 11.62 10.61 9.52 7.65 10.56 K=6
247
+ verify 41: accepted 5/5 (K=6) + correction -> +6 tokens (162.5 ms) cpos=191
248
+ [margins] 12.67 9.39 4.60 9.97 2.02 K=6
249
+ verify 42: accepted 5/5 (K=6) + correction -> +6 tokens (168.0 ms) cpos=197
250
+ [margins] 14.98 7.51 10.44 9.94 7.57 K=6
251
+ verify 43: accepted 5/5 (K=6) + correction -> +6 tokens (164.3 ms) cpos=203
252
+ [margins] 4.09 0.55 3.41 5.00 4.06 K=2
253
+ verify 44: accepted 0/1 (K=2) + correction -> +1 tokens (108.9 ms) cpos=204
254
+ [margins] 5.86 8.56 7.55 6.81 3.27 K=6
255
+ verify 45: accepted 5/5 (K=6) + correction -> +6 tokens (162.2 ms) cpos=210
256
+ [margins] 10.71 10.00 9.00 4.92 9.59 K=6
257
+ verify 46: accepted 5/5 (K=6) + correction -> +6 tokens (169.7 ms) cpos=216
258
+ [margins] 15.51 5.79 7.34 9.90 5.21 K=6
259
+ verify 47: accepted 5/5 (K=6) + correction -> +6 tokens (162.6 ms) cpos=222
260
+ [margins] 14.50 6.66 9.66 5.71 6.84 K=6
261
+ verify 48: accepted 5/5 (K=6) + correction -> +6 tokens (170.3 ms) cpos=228
262
+ [margins] 1.05 8.42 11.61 5.91 1.40 K=5
263
+ verify 49: accepted 4/4 (K=5) + correction -> +5 tokens (147.8 ms) cpos=233
264
+ [margins] 5.88 0.42 3.84 8.93 2.47 K=2
265
+ verify 50: accepted 0/1 (K=2) + correction -> +1 tokens (105.0 ms) cpos=234
266
+
267
+ [spec] generated 200 tokens over 50 verifies: mean tokens/verify = 4.00 (block=5, max 6)
268
+ [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
269
+ [spec] SPEC-DECODE: 34.8 ms/tok = 28.72 tok/s (vs base M=1 68.6 ms/tok = 14.57 tok/s -> 1.97x)
270
+ [decode] PREFILL: 41 positions in 695.5 ms = 59.0 tok/s (16.963 ms/tok)
271
+ [spec] decoding 200 tokens (block=5, draft passes=1, adaptK=1.50, prompt=2 s=42)...
272
+ [margins] 8.22 2.63 0.34 1.30 2.65 K=3
273
+ verify 1: accepted 2/2 (K=3) + correction -> +3 tokens (112.0 ms) cpos=44
274
+ [margins] 5.18 6.08 3.28 2.74 0.40 K=5
275
+ verify 2: accepted 3/4 (K=5) + correction -> +4 tokens (127.8 ms) cpos=48
276
+ [margins] 1.18 8.39 1.93 0.01 1.37 K=4
277
+ verify 3: accepted 0/3 (K=4) + correction -> +1 tokens (123.7 ms) cpos=49
278
+ [margins] 0.53 8.54 2.64 1.66 0.23 K=5
279
+ verify 4: accepted 0/4 (K=5) + correction -> +1 tokens (133.9 ms) cpos=50
280
+ [margins] 5.80 5.51 2.95 1.55 0.59 K=5
281
+ verify 5: accepted 0/4 (K=5) + correction -> +1 tokens (128.1 ms) cpos=51
282
+ [margins] 9.92 5.82 0.93 0.04 0.58 K=3
283
+ verify 6: accepted 1/2 (K=3) + correction -> +2 tokens (111.1 ms) cpos=53
284
+ [margins] 1.80 1.29 8.92 2.67 0.51 K=2
285
+ verify 7: accepted 0/1 (K=2) + correction -> +1 tokens (98.3 ms) cpos=54
286
+ [margins] 2.76 7.15 2.10 2.89 0.93 K=5
287
+ verify 8: accepted 2/4 (K=5) + correction -> +3 tokens (133.5 ms) cpos=57
288
+ [margins] 0.93 5.67 8.81 1.12 0.05 K=4
289
+ verify 9: accepted 0/3 (K=4) + correction -> +1 tokens (127.9 ms) cpos=58
290
+ [margins] 3.08 5.28 1.60 4.38 0.30 K=5
291
+ verify 10: accepted 2/4 (K=5) + correction -> +3 tokens (133.8 ms) cpos=61
292
+ [margins] 4.18 5.29 9.43 2.65 6.32 K=6
293
+ verify 11: accepted 3/5 (K=6) + correction -> +4 tokens (153.4 ms) cpos=65
294
+ [margins] 0.83 6.75 6.03 1.23 1.02 K=4
295
+ verify 12: accepted 3/3 (K=4) + correction -> +4 tokens (129.0 ms) cpos=69
296
+ [margins] 6.50 6.55 3.81 1.33 0.68 K=4
297
+ verify 13: accepted 2/3 (K=4) + correction -> +3 tokens (129.8 ms) cpos=72
298
+ [margins] 0.40 0.81 1.82 4.19 0.82 K=2
299
+ verify 14: accepted 1/1 (K=2) + correction -> +2 tokens (98.4 ms) cpos=74
300
+ [margins] 6.27 3.62 1.84 0.43 4.76 K=4
301
+ verify 15: accepted 1/3 (K=4) + correction -> +2 tokens (126.3 ms) cpos=76
302
+ [margins] 10.32 6.07 5.38 4.26 0.22 K=5
303
+ verify 16: accepted 2/4 (K=5) + correction -> +3 tokens (141.4 ms) cpos=79
304
+ [margins] 15.66 9.45 7.67 5.87 1.59 K=6
305
+ verify 17: accepted 4/5 (K=6) + correction -> +5 tokens (161.3 ms) cpos=84
306
+ [margins] 19.83 11.41 3.61 0.83 10.16 K=4
307
+ verify 18: accepted 2/3 (K=4) + correction -> +3 tokens (131.1 ms) cpos=87
308
+ [margins] 11.27 16.76 13.72 1.38 3.99 K=4
309
+ verify 19: accepted 3/3 (K=4) + correction -> +4 tokens (129.4 ms) cpos=91
310
+ [margins] 15.91 15.07 6.65 4.50 5.16 K=6
311
+ verify 20: accepted 5/5 (K=6) + correction -> +6 tokens (161.5 ms) cpos=97
312
+ [margins] 14.12 16.59 9.39 12.00 4.21 K=6
313
+ verify 21: accepted 5/5 (K=6) + correction -> +6 tokens (158.7 ms) cpos=103
314
+ [margins] 16.71 15.05 10.77 0.88 1.55 K=4
315
+ verify 22: accepted 3/3 (K=4) + correction -> +4 tokens (133.5 ms) cpos=107
316
+ [margins] 14.17 11.77 9.02 3.86 0.70 K=5
317
+ verify 23: accepted 4/4 (K=5) + correction -> +5 tokens (145.8 ms) cpos=112
318
+ [margins] 12.61 8.18 4.79 4.00 0.12 K=5
319
+ verify 24: accepted 2/4 (K=5) + correction -> +3 tokens (143.9 ms) cpos=115
320
+ [margins] 12.37 8.93 10.27 1.41 0.17 K=4
321
+ verify 25: accepted 3/3 (K=4) + correction -> +4 tokens (131.3 ms) cpos=119
322
+ [margins] 5.83 10.85 5.17 4.20 0.96 K=5
323
+ verify 26: accepted 4/4 (K=5) + correction -> +5 tokens (149.3 ms) cpos=124
324
+ [margins] 12.49 12.62 10.44 3.47 0.05 K=5
325
+ verify 27: accepted 4/4 (K=5) + correction -> +5 tokens (157.9 ms) cpos=129
326
+ [margins] 9.96 5.99 10.21 7.86 1.92 K=6
327
+ verify 28: accepted 5/5 (K=6) + correction -> +6 tokens (162.5 ms) cpos=135
328
+ [margins] 16.55 9.48 0.41 5.15 3.59 K=3
329
+ verify 29: accepted 2/2 (K=3) + correction -> +3 tokens (120.2 ms) cpos=138
330
+ [margins] 15.92 7.64 11.74 7.01 2.08 K=6
331
+ verify 30: accepted 4/5 (K=6) + correction -> +5 tokens (166.0 ms) cpos=143
332
+ [margins] 18.00 11.27 8.93 3.38 1.73 K=6
333
+ verify 31: accepted 5/5 (K=6) + correction -> +6 tokens (166.3 ms) cpos=149
334
+ [margins] 12.47 8.98 5.64 0.60 0.53 K=4
335
+ verify 32: accepted 3/3 (K=4) + correction -> +4 tokens (134.0 ms) cpos=153
336
+ [margins] 3.59 10.19 11.00 6.62 5.69 K=6
337
+ verify 33: accepted 4/5 (K=6) + correction -> +5 tokens (161.1 ms) cpos=158
338
+ [margins] 11.96 7.92 9.76 0.15 0.03 K=4
339
+ verify 34: accepted 3/3 (K=4) + correction -> +4 tokens (129.6 ms) cpos=162
340
+ [margins] 14.74 1.44 3.39 4.43 0.59 K=2
341
+ verify 35: accepted 1/1 (K=2) + correction -> +2 tokens (103.6 ms) cpos=164
342
+ [margins] 14.51 9.26 2.88 1.42 1.27 K=4
343
+ verify 36: accepted 2/3 (K=4) + correction -> +3 tokens (133.1 ms) cpos=167
344
+ [margins] 9.72 13.55 13.45 5.82 2.61 K=6
345
+ verify 37: accepted 4/5 (K=6) + correction -> +5 tokens (165.8 ms) cpos=172
346
+ [margins] 16.55 9.27 13.78 7.45 0.74 K=5
347
+ verify 38: accepted 4/4 (K=5) + correction -> +5 tokens (145.5 ms) cpos=177
348
+ [margins] 12.22 15.29 14.33 6.57 3.21 K=6
349
+ verify 39: accepted 5/5 (K=6) + correction -> +6 tokens (160.4 ms) cpos=183
350
+ [margins] 14.43 15.48 9.60 1.62 7.58 K=6
351
+ verify 40: accepted 5/5 (K=6) + correction -> +6 tokens (167.5 ms) cpos=189
352
+ [margins] 20.15 10.37 0.66 11.93 5.61 K=3
353
+ verify 41: accepted 2/2 (K=3) + correction -> +3 tokens (119.2 ms) cpos=192
354
+ [margins] 12.67 8.69 9.34 5.55 2.56 K=6
355
+ verify 42: accepted 5/5 (K=6) + correction -> +6 tokens (163.3 ms) cpos=198
356
+ [margins] 13.45 14.22 8.64 1.15 2.59 K=4
357
+ verify 43: accepted 3/3 (K=4) + correction -> +4 tokens (135.2 ms) cpos=202
358
+ [margins] 14.45 13.55 14.15 6.37 1.53 K=6
359
+ verify 44: accepted 4/5 (K=6) + correction -> +5 tokens (169.6 ms) cpos=207
360
+ [margins] 11.84 8.39 7.54 4.39 4.87 K=6
361
+ verify 45: accepted 5/5 (K=6) + correction -> +6 tokens (162.3 ms) cpos=213
362
+ [margins] 21.74 11.92 7.40 15.82 5.08 K=6
363
+ verify 46: accepted 5/5 (K=6) + correction -> +6 tokens (159.9 ms) cpos=219
364
+ [margins] 20.59 9.04 23.31 5.33 18.86 K=6
365
+ verify 47: accepted 5/5 (K=6) + correction -> +6 tokens (171.0 ms) cpos=225
366
+ [margins] 8.40 26.83 8.00 18.73 8.92 K=6
367
+ verify 48: accepted 5/5 (K=6) + correction -> +6 tokens (163.6 ms) cpos=231
368
+ [margins] 29.37 10.62 15.21 2.19 9.40 K=6
369
+ verify 49: accepted 5/5 (K=6) + correction -> +6 tokens (170.4 ms) cpos=237
370
+ [margins] 11.65 14.80 7.33 1.75 9.92 K=6
371
+ verify 50: accepted 5/5 (K=6) + correction -> +6 tokens (164.1 ms) cpos=243
372
+
373
+ [spec] generated 202 tokens over 50 verifies: mean tokens/verify = 4.04 (block=5, max 6)
374
+ [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
375
+ [spec] SPEC-DECODE: 35.1 ms/tok = 28.45 tok/s (vs base M=1 68.6 ms/tok = 14.57 tok/s -> 1.95x)
376
+ [decode] PREFILL: 26 positions in 455.1 ms = 57.1 tok/s (17.503 ms/tok)
377
+ [spec] decoding 200 tokens (block=5, draft passes=1, adaptK=1.50, prompt=3 s=27)...
378
+ [margins] 1.48 4.67 1.34 0.92 0.66 K=3
379
+ verify 1: accepted 1/2 (K=3) + correction -> +2 tokens (115.3 ms) cpos=28
380
+ [margins] 5.28 0.64 0.42 0.42 0.36 K=2
381
+ verify 2: accepted 1/1 (K=2) + correction -> +2 tokens (100.2 ms) cpos=30
382
+ [margins] 4.37 0.37 0.82 0.80 6.65 K=2
383
+ verify 3: accepted 1/1 (K=2) + correction -> +2 tokens (103.5 ms) cpos=32
384
+ [margins] 3.74 1.34 2.58 0.18 1.50 K=2
385
+ verify 4: accepted 1/1 (K=2) + correction -> +2 tokens (98.8 ms) cpos=34
386
+ [margins] 4.66 2.96 1.29 1.07 0.41 K=3
387
+ verify 5: accepted 2/2 (K=3) + correction -> +3 tokens (110.5 ms) cpos=37
388
+ [margins] 4.70 7.23 5.68 1.81 0.32 K=5
389
+ verify 6: accepted 4/4 (K=5) + correction -> +5 tokens (129.1 ms) cpos=42
390
+ [margins] 1.74 8.87 1.87 4.32 1.29 K=5
391
+ verify 7: accepted 4/4 (K=5) + correction -> +5 tokens (128.9 ms) cpos=47
392
+ [margins] 6.95 5.51 8.57 1.42 4.51 K=4
393
+ verify 8: accepted 3/3 (K=4) + correction -> +4 tokens (120.6 ms) cpos=51
394
+ [margins] 6.71 4.14 8.97 1.53 4.32 K=6
395
+ verify 9: accepted 5/5 (K=6) + correction -> +6 tokens (151.2 ms) cpos=57
396
+ [margins] 7.84 5.56 8.66 1.41 4.57 K=4
397
+ verify 10: accepted 3/3 (K=4) + correction -> +4 tokens (121.3 ms) cpos=61
398
+ [margins] 7.09 3.00 8.23 1.46 5.19 K=4
399
+ verify 11: accepted 3/3 (K=4) + correction -> +4 tokens (120.6 ms) cpos=65
400
+ [margins] 6.68 4.67 8.26 1.18 4.55 K=4
401
+ verify 12: accepted 3/3 (K=4) + correction -> +4 tokens (119.9 ms) cpos=69
402
+ [margins] 7.60 2.78 8.19 1.39 4.99 K=4
403
+ verify 13: accepted 3/3 (K=4) + correction -> +4 tokens (120.6 ms) cpos=73
404
+ [margins] 7.90 1.72 7.82 2.60 5.58 K=6
405
+ verify 14: accepted 1/5 (K=6) + correction -> +2 tokens (147.4 ms) cpos=75
406
+ [margins] 9.23 0.45 0.67 4.80 1.69 K=2
407
+ verify 15: accepted 0/1 (K=2) + correction -> +1 tokens (102.0 ms) cpos=76
408
+ [margins] 1.39 4.11 0.82 0.58 0.15 K=3
409
+ verify 16: accepted 0/2 (K=3) + correction -> +1 tokens (110.9 ms) cpos=77
410
+ [margins] 2.31 0.86 2.14 1.33 1.36 K=2
411
+ verify 17: accepted 0/1 (K=2) + correction -> +1 tokens (98.5 ms) cpos=78
412
+ [margins] 1.54 3.03 3.24 2.93 2.04 K=6
413
+ verify 18: accepted 2/5 (K=6) + correction -> +3 tokens (160.6 ms) cpos=81
414
+ [margins] 2.59 0.68 0.04 1.67 1.40 K=2
415
+ verify 19: accepted 1/1 (K=2) + correction -> +2 tokens (97.8 ms) cpos=83
416
+ [margins] 0.51 0.57 0.60 0.47 0.43 K=2
417
+ verify 20: accepted 1/1 (K=2) + correction -> +2 tokens (101.4 ms) cpos=85
418
+ [margins] 5.35 1.94 0.58 0.99 1.56 K=3
419
+ verify 21: accepted 1/2 (K=3) + correction -> +2 tokens (112.7 ms) cpos=87
420
+ [margins] 4.34 3.29 0.97 3.35 0.71 K=3
421
+ verify 22: accepted 1/2 (K=3) + correction -> +2 tokens (116.0 ms) cpos=89
422
+ [margins] 0.74 2.83 0.40 1.13 0.09 K=3
423
+ verify 23: accepted 2/2 (K=3) + correction -> +3 tokens (114.2 ms) cpos=92
424
+ [margins] 4.53 1.56 0.18 2.32 2.59 K=3
425
+ verify 24: accepted 2/2 (K=3) + correction -> +3 tokens (111.1 ms) cpos=95
426
+ [margins] 7.04 0.54 3.45 1.82 1.80 K=2
427
+ verify 25: accepted 1/1 (K=2) + correction -> +2 tokens (102.4 ms) cpos=97
428
+ [margins] 3.18 2.49 0.52 3.20 1.20 K=3
429
+ verify 26: accepted 2/2 (K=3) + correction -> +3 tokens (117.8 ms) cpos=100
430
+ [margins] 0.10 1.82 1.51 0.80 1.98 K=4
431
+ verify 27: accepted 1/3 (K=4) + correction -> +2 tokens (128.3 ms) cpos=102
432
+ [margins] 0.51 0.84 0.75 3.34 7.21 K=2
433
+ verify 28: accepted 0/1 (K=2) + correction -> +1 tokens (102.4 ms) cpos=103
434
+ [margins] 6.52 6.34 0.99 3.69 7.39 K=3
435
+ verify 29: accepted 2/2 (K=3) + correction -> +3 tokens (112.6 ms) cpos=106
436
+ [margins] 7.87 5.76 1.14 4.40 2.20 K=3
437
+ verify 30: accepted 2/2 (K=3) + correction -> +3 tokens (114.0 ms) cpos=109
438
+ [margins] 9.39 1.82 1.85 1.33 1.89 K=4
439
+ verify 31: accepted 3/3 (K=4) + correction -> +4 tokens (129.1 ms) cpos=113
440
+ [margins] 0.71 3.44 2.32 2.12 3.54 K=6
441
+ verify 32: accepted 0/5 (K=6) + correction -> +1 tokens (156.3 ms) cpos=114
442
+ [margins] 0.03 9.06 1.76 3.99 2.83 K=6
443
+ verify 33: accepted 2/5 (K=6) + correction -> +3 tokens (162.1 ms) cpos=117
444
+ [margins] 1.43 5.68 2.98 3.61 8.69 K=6
445
+ verify 34: accepted 5/5 (K=6) + correction -> +6 tokens (161.6 ms) cpos=123
446
+ [margins] 10.31 12.19 4.68 12.29 5.24 K=6
447
+ verify 35: accepted 5/5 (K=6) + correction -> +6 tokens (178.0 ms) cpos=129
448
+ [margins] 11.92 11.45 12.60 10.78 2.81 K=6
449
+ verify 36: accepted 5/5 (K=6) + correction -> +6 tokens (160.2 ms) cpos=135
450
+ [margins] 3.17 0.88 0.55 0.00 1.50 K=2
451
+ verify 37: accepted 1/1 (K=2) + correction -> +2 tokens (107.6 ms) cpos=137
452
+ [margins] 1.00 3.34 1.20 0.60 0.15 K=3
453
+ verify 38: accepted 0/2 (K=3) + correction -> +1 tokens (116.3 ms) cpos=138
454
+ [margins] 0.09 1.30 0.77 0.50 0.06 K=2
455
+ verify 39: accepted 1/1 (K=2) + correction -> +2 tokens (104.1 ms) cpos=140
456
+ [margins] 3.95 1.59 0.59 2.01 2.45 K=3
457
+ verify 40: accepted 0/2 (K=3) + correction -> +1 tokens (113.0 ms) cpos=141
458
+ [margins] 1.76 0.19 2.99 0.68 1.56 K=2
459
+ verify 41: accepted 1/1 (K=2) + correction -> +2 tokens (100.0 ms) cpos=143
460
+ [margins] 0.20 0.12 0.76 0.69 0.37 K=2
461
+ verify 42: accepted 0/1 (K=2) + correction -> +1 tokens (104.8 ms) cpos=144
462
+ [margins] 1.48 3.67 1.67 2.82 0.37 K=5
463
+ verify 43: accepted 1/4 (K=5) + correction -> +2 tokens (141.9 ms) cpos=146
464
+ [margins] 5.60 0.54 0.83 0.56 0.54 K=2
465
+ verify 44: accepted 1/1 (K=2) + correction -> +2 tokens (104.1 ms) cpos=148
466
+ [margins] 0.89 0.22 0.10 2.79 1.69 K=2
467
+ verify 45: accepted 1/1 (K=2) + correction -> +2 tokens (99.8 ms) cpos=150
468
+ [margins] 1.50 0.51 0.93 0.73 1.27 K=2
469
+ verify 46: accepted 1/1 (K=2) + correction -> +2 tokens (103.1 ms) cpos=152
470
+ [margins] 3.66 0.10 1.36 2.19 1.17 K=2
471
+ verify 47: accepted 1/1 (K=2) + correction -> +2 tokens (99.7 ms) cpos=154
472
+ [margins] 0.60 2.28 3.00 0.18 0.39 K=4
473
+ verify 48: accepted 1/3 (K=4) + correction -> +2 tokens (129.9 ms) cpos=156
474
+ [margins] 5.71 2.16 0.50 0.90 0.06 K=3
475
+ verify 49: accepted 1/2 (K=3) + correction -> +2 tokens (111.2 ms) cpos=158
476
+ [margins] 3.47 2.34 1.39 0.96 0.96 K=3
477
+ verify 50: accepted 2/2 (K=3) + correction -> +3 tokens (118.8 ms) cpos=161
478
+ [margins] 8.51 3.33 2.63 1.10 0.99 K=4
479
+ verify 51: accepted 1/3 (K=4) + correction -> +2 tokens (128.2 ms) cpos=163
480
+ [margins] 1.81 0.77 0.15 1.68 1.83 K=2
481
+ verify 52: accepted 1/1 (K=2) + correction -> +2 tokens (103.2 ms) cpos=165
482
+ [margins] 3.12 3.98 2.16 2.02 2.93 K=6
483
+ verify 53: accepted 2/5 (K=6) + correction -> +3 tokens (157.7 ms) cpos=168
484
+ [margins] 4.96 1.61 6.24 2.24 0.05 K=5
485
+ verify 54: accepted 4/4 (K=5) + correction -> +5 tokens (144.8 ms) cpos=173
486
+ [margins] 0.19 7.13 0.59 0.17 0.25 K=3
487
+ verify 55: accepted 0/2 (K=3) + correction -> +1 tokens (116.6 ms) cpos=174
488
+ [margins] 0.09 0.21 3.36 0.47 4.10 K=2
489
+ verify 56: accepted 0/1 (K=2) + correction -> +1 tokens (104.9 ms) cpos=175
490
+ [margins] 0.12 0.96 3.10 0.45 1.62 K=2
491
+ verify 57: accepted 1/1 (K=2) + correction -> +2 tokens (107.3 ms) cpos=177
492
+ [margins] 3.01 3.26 3.16 1.47 2.84 K=4
493
+ verify 58: accepted 2/3 (K=4) + correction -> +3 tokens (128.5 ms) cpos=180
494
+ [margins] 0.81 0.23 0.71 1.46 0.79 K=2
495
+ verify 59: accepted 0/1 (K=2) + correction -> +1 tokens (100.5 ms) cpos=181
496
+ [margins] 5.66 0.40 0.02 3.27 7.62 K=2
497
+ verify 60: accepted 1/1 (K=2) + correction -> +2 tokens (100.6 ms) cpos=183
498
+ [margins] 3.42 4.08 5.21 3.40 1.90 K=6
499
+ verify 61: accepted 2/5 (K=6) + correction -> +3 tokens (160.9 ms) cpos=186
500
+ [margins] 3.22 2.73 0.78 0.78 0.79 K=3
501
+ verify 62: accepted 2/2 (K=3) + correction -> +3 tokens (118.1 ms) cpos=189
502
+ [margins] 4.45 1.61 0.96 0.71 0.21 K=3
503
+ verify 63: accepted 2/2 (K=3) + correction -> +3 tokens (120.1 ms) cpos=192
504
+ [margins] 5.56 0.89 0.02 0.55 0.88 K=2
505
+ verify 64: accepted 1/1 (K=2) + correction -> +2 tokens (100.8 ms) cpos=194
506
+ [margins] 0.15 0.60 2.98 1.84 0.14 K=2
507
+ verify 65: accepted 0/1 (K=2) + correction -> +1 tokens (106.0 ms) cpos=195
508
+ [margins] 2.55 0.52 1.33 4.17 0.70 K=2
509
+ verify 66: accepted 1/1 (K=2) + correction -> +2 tokens (105.3 ms) cpos=197
510
+ [margins] 0.60 1.75 6.71 0.55 3.24 K=4
511
+ verify 67: accepted 0/3 (K=4) + correction -> +1 tokens (130.5 ms) cpos=198
512
+ [margins] 2.72 0.86 1.49 4.63 0.86 K=2
513
+ verify 68: accepted 0/1 (K=2) + correction -> +1 tokens (106.3 ms) cpos=199
514
+ [margins] 5.05 8.27 7.55 2.19 5.42 K=6
515
+ verify 69: accepted 0/5 (K=6) + correction -> +1 tokens (166.4 ms) cpos=200
516
+ [margins] 8.66 6.20 0.27 3.78 2.16 K=3
517
+ verify 70: accepted 2/2 (K=3) + correction -> +3 tokens (112.8 ms) cpos=203
518
+ [margins] 4.11 4.18 2.15 1.22 0.60 K=4
519
+ verify 71: accepted 1/3 (K=4) + correction -> +2 tokens (128.2 ms) cpos=205
520
+ [margins] 1.59 0.62 0.11 0.97 2.01 K=2
521
+ verify 72: accepted 1/1 (K=2) + correction -> +2 tokens (100.4 ms) cpos=207
522
+ [margins] 0.50 1.37 0.65 3.45 2.94 K=2
523
+ verify 73: accepted 0/1 (K=2) + correction -> +1 tokens (104.3 ms) cpos=208
524
+ [margins] 10.96 1.27 1.43 4.93 1.36 K=2
525
+ verify 74: accepted 1/1 (K=2) + correction -> +2 tokens (101.5 ms) cpos=210
526
+ [margins] 2.46 2.68 0.96 0.77 3.83 K=3
527
+ verify 75: accepted 2/2 (K=3) + correction -> +3 tokens (119.6 ms) cpos=213
528
+ [margins] 1.26 0.13 0.11 1.21 0.36 K=2
529
+ verify 76: accepted 1/1 (K=2) + correction -> +2 tokens (99.3 ms) cpos=215
530
+ [margins] 3.45 3.22 7.61 5.54 0.22 K=5
531
+ verify 77: accepted 4/4 (K=5) + correction -> +5 tokens (150.1 ms) cpos=220
532
+ [margins] 3.92 2.51 2.97 0.15 3.78 K=4
533
+ verify 78: accepted 0/3 (K=4) + correction -> +1 tokens (133.9 ms) cpos=221
534
+ [margins] 11.65 2.78 2.67 1.08 4.60 K=4
535
+ verify 79: accepted 3/3 (K=4) + correction -> +4 tokens (131.5 ms) cpos=225
536
+ [margins] 12.17 4.09 0.34 1.69 0.33 K=3
537
+ verify 80: accepted 2/2 (K=3) + correction -> +3 tokens (122.6 ms) cpos=228
538
+
539
+ [spec] generated 202 tokens over 80 verifies: mean tokens/verify = 2.52 (block=5, max 6)
540
+ [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
541
+ [spec] SPEC-DECODE: 47.4 ms/tok = 21.08 tok/s (vs base M=1 68.6 ms/tok = 14.57 tok/s -> 1.45x)
542
+ [decode] PREFILL: 14 positions in 284.9 ms = 49.1 tok/s (20.348 ms/tok)
543
+ [spec] decoding 200 tokens (block=5, draft passes=1, adaptK=1.50, prompt=4 s=15)...
544
+ [margins] 0.01 8.63 2.90 3.16 2.74 K=6
545
+ verify 1: accepted 0/5 (K=6) + correction -> +1 tokens (158.9 ms) cpos=15
546
+ [margins] 0.38 1.46 0.23 0.02 0.23 K=2
547
+ verify 2: accepted 1/1 (K=2) + correction -> +2 tokens (101.1 ms) cpos=17
548
+ [margins] 0.34 0.29 0.30 1.05 0.82 K=2
549
+ verify 3: accepted 0/1 (K=2) + correction -> +1 tokens (96.3 ms) cpos=18
550
+ [margins] 0.78 0.66 0.20 0.63 0.35 K=2
551
+ verify 4: accepted 1/1 (K=2) + correction -> +2 tokens (99.9 ms) cpos=20
552
+ [margins] 0.85 0.78 0.02 0.26 0.25 K=2
553
+ verify 5: accepted 1/1 (K=2) + correction -> +2 tokens (95.0 ms) cpos=22
554
+ [margins] 2.01 1.20 0.42 0.50 1.42 K=2
555
+ verify 6: accepted 1/1 (K=2) + correction -> +2 tokens (101.2 ms) cpos=24
556
+ [margins] 0.21 1.28 0.61 0.05 3.77 K=2
557
+ verify 7: accepted 0/1 (K=2) + correction -> +1 tokens (96.7 ms) cpos=25
558
+ [margins] 0.44 0.67 0.61 0.59 0.60 K=2
559
+ verify 8: accepted 1/1 (K=2) + correction -> +2 tokens (96.7 ms) cpos=27
560
+ [margins] 0.30 0.49 0.32 0.77 1.55 K=2
561
+ verify 9: accepted 1/1 (K=2) + correction -> +2 tokens (104.3 ms) cpos=29
562
+ [margins] 3.74 1.80 0.16 2.43 0.04 K=3
563
+ verify 10: accepted 2/2 (K=3) + correction -> +3 tokens (113.0 ms) cpos=32
564
+ [margins] 0.78 0.35 0.09 0.00 1.15 K=2
565
+ verify 11: accepted 1/1 (K=2) + correction -> +2 tokens (95.4 ms) cpos=34
566
+ [margins] 1.93 1.67 1.38 1.19 0.44 K=3
567
+ verify 12: accepted 1/2 (K=3) + correction -> +2 tokens (110.1 ms) cpos=36
568
+ [margins] 0.96 0.13 1.03 1.39 1.05 K=2
569
+ verify 13: accepted 0/1 (K=2) + correction -> +1 tokens (96.1 ms) cpos=37
570
+ [margins] 0.26 2.39 0.85 0.22 0.21 K=3
571
+ verify 14: accepted 1/2 (K=3) + correction -> +2 tokens (113.3 ms) cpos=39
572
+ [margins] 3.81 1.22 0.52 0.44 1.64 K=2
573
+ verify 15: accepted 0/1 (K=2) + correction -> +1 tokens (99.2 ms) cpos=40
574
+ [margins] 3.25 3.88 0.94 2.44 1.58 K=3
575
+ verify 16: accepted 1/2 (K=3) + correction -> +2 tokens (110.1 ms) cpos=42
576
+ [margins] 0.27 0.04 0.75 0.91 0.08 K=2
577
+ verify 17: accepted 1/1 (K=2) + correction -> +2 tokens (101.8 ms) cpos=44
578
+ [margins] 1.27 7.42 0.11 1.32 1.84 K=3
579
+ verify 18: accepted 2/2 (K=3) + correction -> +3 tokens (110.5 ms) cpos=47
580
+ [margins] 3.87 7.83 2.27 0.88 2.79 K=4
581
+ verify 19: accepted 3/3 (K=4) + correction -> +4 tokens (125.9 ms) cpos=51
582
+ [margins] 2.79 0.31 0.36 0.21 2.84 K=2
583
+ verify 20: accepted 0/1 (K=2) + correction -> +1 tokens (102.3 ms) cpos=52
584
+ [margins] 2.06 1.29 1.72 1.56 3.65 K=2
585
+ verify 21: accepted 0/1 (K=2) + correction -> +1 tokens (98.9 ms) cpos=53
586
+ [margins] 8.37 3.11 1.18 0.62 2.69 K=3
587
+ verify 22: accepted 2/2 (K=3) + correction -> +3 tokens (117.2 ms) cpos=56
588
+ [margins] 5.15 1.91 3.42 3.45 0.61 K=5
589
+ verify 23: accepted 4/4 (K=5) + correction -> +5 tokens (137.3 ms) cpos=61
590
+ [margins] 0.01 1.86 0.93 3.17 1.39 K=3
591
+ verify 24: accepted 0/2 (K=3) + correction -> +1 tokens (114.1 ms) cpos=62
592
+ [margins] 1.83 1.00 1.57 1.95 2.58 K=2
593
+ verify 25: accepted 0/1 (K=2) + correction -> +1 tokens (100.4 ms) cpos=63
594
+ [margins] 1.01 1.37 1.08 0.54 0.45 K=2
595
+ verify 26: accepted 1/1 (K=2) + correction -> +2 tokens (102.0 ms) cpos=65
596
+ [margins] 0.31 2.24 0.50 0.93 1.04 K=3
597
+ verify 27: accepted 0/2 (K=3) + correction -> +1 tokens (113.6 ms) cpos=66
598
+ [margins] 0.22 4.94 0.72 0.11 0.42 K=3
599
+ verify 28: accepted 0/2 (K=3) + correction -> +1 tokens (114.2 ms) cpos=67
600
+ [margins] 1.72 0.19 0.00 0.29 2.26 K=2
601
+ verify 29: accepted 1/1 (K=2) + correction -> +2 tokens (104.4 ms) cpos=69
602
+ [margins] 0.67 2.48 0.67 1.78 0.07 K=3
603
+ verify 30: accepted 0/2 (K=3) + correction -> +1 tokens (115.8 ms) cpos=70
604
+ [margins] 1.29 1.17 0.40 0.97 0.11 K=2
605
+ verify 31: accepted 0/1 (K=2) + correction -> +1 tokens (102.2 ms) cpos=71
606
+ [margins] 0.12 0.34 0.33 0.82 1.47 K=2
607
+ verify 32: accepted 0/1 (K=2) + correction -> +1 tokens (103.6 ms) cpos=72
608
+ [margins] 0.13 0.12 1.61 0.63 0.38 K=2
609
+ verify 33: accepted 0/1 (K=2) + correction -> +1 tokens (98.1 ms) cpos=73
610
+ [margins] 7.59 1.80 1.05 0.75 0.20 K=3
611
+ verify 34: accepted 2/2 (K=3) + correction -> +3 tokens (113.7 ms) cpos=76
612
+ [margins] 0.17 0.75 2.36 0.74 1.64 K=2
613
+ verify 35: accepted 1/1 (K=2) + correction -> +2 tokens (97.9 ms) cpos=78
614
+ [margins] 0.34 0.06 2.24 0.52 9.83 K=2
615
+ verify 36: accepted 1/1 (K=2) + correction -> +2 tokens (104.4 ms) cpos=80
616
+ [margins] 1.54 4.99 2.21 2.38 9.55 K=6
617
+ verify 37: accepted 5/5 (K=6) + correction -> +6 tokens (163.0 ms) cpos=86
618
+ [margins] 7.12 8.22 4.56 4.00 2.24 K=6
619
+ verify 38: accepted 2/5 (K=6) + correction -> +3 tokens (151.2 ms) cpos=89
620
+ [margins] 9.38 5.33 2.94 1.31 0.32 K=4
621
+ verify 39: accepted 1/3 (K=4) + correction -> +2 tokens (123.0 ms) cpos=91
622
+ [margins] 7.12 4.27 2.56 0.90 1.80 K=4
623
+ verify 40: accepted 1/3 (K=4) + correction -> +2 tokens (122.4 ms) cpos=93
624
+ [margins] 5.38 0.03 3.37 0.46 0.16 K=2
625
+ verify 41: accepted 1/1 (K=2) + correction -> +2 tokens (101.5 ms) cpos=95
626
+ [margins] 5.08 2.94 2.05 1.17 4.67 K=4
627
+ verify 42: accepted 3/3 (K=4) + correction -> +4 tokens (122.2 ms) cpos=99
628
+ [margins] 3.12 6.31 2.86 2.76 2.25 K=6
629
+ verify 43: accepted 5/5 (K=6) + correction -> +6 tokens (153.2 ms) cpos=105
630
+ [margins] 2.71 8.53 2.55 5.11 2.84 K=6
631
+ verify 44: accepted 5/5 (K=6) + correction -> +6 tokens (148.7 ms) cpos=111
632
+ [margins] 2.92 9.11 2.82 4.95 3.12 K=6
633
+ verify 45: accepted 5/5 (K=6) + correction -> +6 tokens (154.8 ms) cpos=117
634
+ [margins] 3.45 8.54 2.84 4.46 2.94 K=6
635
+ verify 46: accepted 5/5 (K=6) + correction -> +6 tokens (151.1 ms) cpos=123
636
+ [margins] 3.62 8.56 2.74 5.23 2.78 K=6
637
+ verify 47: accepted 5/5 (K=6) + correction -> +6 tokens (170.2 ms) cpos=129
638
+ [margins] 3.89 9.11 2.96 5.21 2.52 K=6
639
+ verify 48: accepted 5/5 (K=6) + correction -> +6 tokens (151.2 ms) cpos=135
640
+ [margins] 3.10 9.44 1.76 4.55 2.04 K=6
641
+ verify 49: accepted 5/5 (K=6) + correction -> +6 tokens (155.3 ms) cpos=141
642
+ [margins] 4.30 9.53 3.36 4.49 3.08 K=6
643
+ verify 50: accepted 5/5 (K=6) + correction -> +6 tokens (151.4 ms) cpos=147
644
+ [margins] 4.50 9.32 2.88 4.76 2.59 K=6
645
+ verify 51: accepted 5/5 (K=6) + correction -> +6 tokens (157.6 ms) cpos=153
646
+ [margins] 2.96 10.21 1.40 6.13 2.34 K=3
647
+ verify 52: accepted 2/2 (K=3) + correction -> +3 tokens (118.0 ms) cpos=156
648
+ [margins] 6.38 1.34 5.25 1.66 2.31 K=2
649
+ verify 53: accepted 1/1 (K=2) + correction -> +2 tokens (100.8 ms) cpos=158
650
+ [margins] 4.58 0.54 0.24 0.53 0.94 K=2
651
+ verify 54: accepted 0/1 (K=2) + correction -> +1 tokens (104.9 ms) cpos=159
652
+ [margins] 8.20 0.39 0.26 0.16 1.51 K=2
653
+ verify 55: accepted 1/1 (K=2) + correction -> +2 tokens (105.4 ms) cpos=161
654
+ [margins] 9.93 3.58 1.58 4.13 2.47 K=6
655
+ verify 56: accepted 4/5 (K=6) + correction -> +5 tokens (157.2 ms) cpos=166
656
+ [margins] 2.08 3.04 4.06 0.32 1.21 K=4
657
+ verify 57: accepted 0/3 (K=4) + correction -> +1 tokens (132.1 ms) cpos=167
658
+ [margins] 9.36 4.18 7.96 7.27 4.74 K=6
659
+ verify 58: accepted 5/5 (K=6) + correction -> +6 tokens (166.4 ms) cpos=173
660
+ [margins] 6.12 0.52 0.57 2.33 0.32 K=2
661
+ verify 59: accepted 1/1 (K=2) + correction -> +2 tokens (99.6 ms) cpos=175
662
+ [margins] 6.66 7.54 4.32 1.80 3.22 K=6
663
+ verify 60: accepted 5/5 (K=6) + correction -> +6 tokens (167.6 ms) cpos=181
664
+ [margins] 11.55 10.84 2.76 2.52 4.00 K=6
665
+ verify 61: accepted 5/5 (K=6) + correction -> +6 tokens (163.1 ms) cpos=187
666
+ [margins] 20.48 6.85 3.96 3.78 3.58 K=6
667
+ verify 62: accepted 4/5 (K=6) + correction -> +5 tokens (169.0 ms) cpos=192
668
+ [margins] 7.93 7.46 1.88 8.20 9.99 K=6
669
+ verify 63: accepted 5/5 (K=6) + correction -> +6 tokens (162.5 ms) cpos=198
670
+ [margins] 13.73 13.65 6.14 3.87 10.56 K=6
671
+ verify 64: accepted 5/5 (K=6) + correction -> +6 tokens (170.8 ms) cpos=204
672
+ [margins] 22.41 8.11 5.10 7.31 6.49 K=6
673
+ verify 65: accepted 5/5 (K=6) + correction -> +6 tokens (159.9 ms) cpos=210
674
+ [margins] 13.18 12.17 16.55 10.06 9.45 K=6
675
+ verify 66: accepted 5/5 (K=6) + correction -> +6 tokens (167.1 ms) cpos=216
676
+
677
+ [spec] generated 202 tokens over 66 verifies: mean tokens/verify = 3.06 (block=5, max 6)
678
+ [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
679
+ [spec] SPEC-DECODE: 39.9 ms/tok = 25.04 tok/s (vs base M=1 68.6 ms/tok = 14.57 tok/s -> 1.72x)
680
+ [decode] PREFILL: 82 positions in 1325.3 ms = 61.9 tok/s (16.162 ms/tok)
681
+ [spec] decoding 200 tokens (block=5, draft passes=1, adaptK=1.50, prompt=5 s=83)...
682
+ [margins] 6.90 1.07 0.12 1.48 1.43 K=2
683
+ verify 1: accepted 1/1 (K=2) + correction -> +2 tokens (107.2 ms) cpos=84
684
+ [margins] 0.70 0.16 0.55 1.23 0.03 K=2
685
+ verify 2: accepted 0/1 (K=2) + correction -> +1 tokens (101.2 ms) cpos=85
686
+ [margins] 0.17 0.01 0.60 2.16 0.30 K=2
687
+ verify 3: accepted 1/1 (K=2) + correction -> +2 tokens (99.4 ms) cpos=87
688
+ [margins] 3.40 2.99 3.04 1.38 1.69 K=4
689
+ verify 4: accepted 3/3 (K=4) + correction -> +4 tokens (129.4 ms) cpos=91
690
+ [margins] 13.64 6.23 5.25 5.03 4.14 K=6
691
+ verify 5: accepted 5/5 (K=6) + correction -> +6 tokens (164.4 ms) cpos=97
692
+ [margins] 8.06 1.18 2.06 0.72 2.18 K=2
693
+ verify 6: accepted 1/1 (K=2) + correction -> +2 tokens (100.3 ms) cpos=99
694
+ [margins] 10.22 1.24 3.08 2.16 1.06 K=2
695
+ verify 7: accepted 1/1 (K=2) + correction -> +2 tokens (105.1 ms) cpos=101
696
+ [margins] 3.95 0.39 0.03 2.02 1.98 K=2
697
+ verify 8: accepted 1/1 (K=2) + correction -> +2 tokens (102.8 ms) cpos=103
698
+ [margins] 2.34 0.89 2.43 0.26 1.98 K=2
699
+ verify 9: accepted 1/1 (K=2) + correction -> +2 tokens (103.9 ms) cpos=105
700
+ [margins] 9.41 8.33 8.50 4.76 1.84 K=6
701
+ verify 10: accepted 5/5 (K=6) + correction -> +6 tokens (159.9 ms) cpos=111
702
+ [margins] 7.30 7.99 7.40 10.61 1.87 K=6
703
+ verify 11: accepted 4/5 (K=6) + correction -> +5 tokens (168.0 ms) cpos=116
704
+ [margins] 11.85 14.75 7.42 2.01 1.51 K=6
705
+ verify 12: accepted 5/5 (K=6) + correction -> +6 tokens (162.1 ms) cpos=122
706
+ [margins] 8.90 4.92 7.48 1.89 2.90 K=6
707
+ verify 13: accepted 3/5 (K=6) + correction -> +4 tokens (181.6 ms) cpos=126
708
+ [margins] 12.65 2.92 2.82 12.71 6.47 K=6
709
+ verify 14: accepted 5/5 (K=6) + correction -> +6 tokens (181.4 ms) cpos=132
710
+ [margins] 12.77 15.11 0.65 10.58 12.73 K=3
711
+ verify 15: accepted 2/2 (K=3) + correction -> +3 tokens (115.5 ms) cpos=135
712
+ [margins] 15.28 15.66 11.14 8.80 2.59 K=6
713
+ verify 16: accepted 4/5 (K=6) + correction -> +5 tokens (167.2 ms) cpos=140
714
+ [margins] 5.05 3.67 6.01 4.74 5.21 K=6
715
+ verify 17: accepted 0/5 (K=6) + correction -> +1 tokens (156.3 ms) cpos=141
716
+ [margins] 8.74 7.32 0.99 3.14 0.27 K=3
717
+ verify 18: accepted 2/2 (K=3) + correction -> +3 tokens (118.1 ms) cpos=144
718
+ [margins] 10.88 5.31 4.68 1.95 2.16 K=6
719
+ verify 19: accepted 4/5 (K=6) + correction -> +5 tokens (162.1 ms) cpos=149
720
+ [margins] 15.53 10.57 1.85 6.49 11.66 K=6
721
+ verify 20: accepted 5/5 (K=6) + correction -> +6 tokens (161.5 ms) cpos=155
722
+ [margins] 17.17 12.11 13.38 9.05 0.53 K=5
723
+ verify 21: accepted 4/4 (K=5) + correction -> +5 tokens (149.2 ms) cpos=160
724
+ [margins] 10.42 2.59 9.22 6.81 4.29 K=6
725
+ verify 22: accepted 0/5 (K=6) + correction -> +1 tokens (157.2 ms) cpos=161
726
+ [margins] 11.52 6.07 0.85 5.75 0.97 K=3
727
+ verify 23: accepted 2/2 (K=3) + correction -> +3 tokens (117.2 ms) cpos=164
728
+ [margins] 10.45 8.09 5.55 8.29 8.22 K=6
729
+ verify 24: accepted 5/5 (K=6) + correction -> +6 tokens (161.9 ms) cpos=170
730
+ [margins] 1.38 7.30 11.27 6.31 13.32 K=6
731
+ verify 25: accepted 0/5 (K=6) + correction -> +1 tokens (166.9 ms) cpos=171
732
+ [margins] 0.31 8.92 10.72 5.05 8.66 K=6
733
+ verify 26: accepted 3/5 (K=6) + correction -> +4 tokens (168.3 ms) cpos=175
734
+ [margins] 16.55 17.06 13.01 7.74 1.49 K=5
735
+ verify 27: accepted 4/4 (K=5) + correction -> +5 tokens (149.7 ms) cpos=180
736
+ [margins] 17.17 16.06 6.59 6.39 7.71 K=6
737
+ verify 28: accepted 5/5 (K=6) + correction -> +6 tokens (160.9 ms) cpos=186
738
+ [margins] 8.06 8.29 2.92 7.21 2.80 K=6
739
+ verify 29: accepted 5/5 (K=6) + correction -> +6 tokens (167.5 ms) cpos=192
740
+ [margins] 14.01 16.08 8.93 11.02 13.03 K=6
741
+ verify 30: accepted 5/5 (K=6) + correction -> +6 tokens (161.9 ms) cpos=198
742
+ [margins] 9.55 9.30 13.07 14.86 3.76 K=6
743
+ verify 31: accepted 5/5 (K=6) + correction -> +6 tokens (164.1 ms) cpos=204
744
+ [margins] 9.72 2.69 5.07 6.21 5.18 K=6
745
+ verify 32: accepted 5/5 (K=6) + correction -> +6 tokens (161.6 ms) cpos=210
746
+ [margins] 5.90 7.11 6.97 9.63 0.50 K=5
747
+ verify 33: accepted 4/4 (K=5) + correction -> +5 tokens (145.3 ms) cpos=215
748
+ [margins] 16.32 18.20 13.04 10.79 0.15 K=5
749
+ verify 34: accepted 4/4 (K=5) + correction -> +5 tokens (151.9 ms) cpos=220
750
+ [margins] 19.59 14.11 4.53 4.83 5.60 K=6
751
+ verify 35: accepted 5/5 (K=6) + correction -> +6 tokens (159.7 ms) cpos=226
752
+ [margins] 8.69 5.83 2.58 11.91 1.02 K=5
753
+ verify 36: accepted 4/4 (K=5) + correction -> +5 tokens (146.6 ms) cpos=231
754
+ [margins] 5.56 13.43 5.46 2.89 8.80 K=6
755
+ verify 37: accepted 3/5 (K=6) + correction -> +4 tokens (169.1 ms) cpos=235
756
+ [margins] 7.08 14.35 9.60 6.23 1.06 K=5
757
+ verify 38: accepted 4/4 (K=5) + correction -> +5 tokens (152.8 ms) cpos=240
758
+ [margins] 20.42 15.29 5.80 5.10 5.25 K=6
759
+ verify 39: accepted 5/5 (K=6) + correction -> +6 tokens (162.7 ms) cpos=246
760
+ [margins] 10.58 6.30 2.62 6.58 10.12 K=6
761
+ verify 40: accepted 2/5 (K=6) + correction -> +3 tokens (169.2 ms) cpos=249
762
+ [margins] 14.35 9.28 3.44 11.69 11.06 K=6
763
+ verify 41: accepted 5/5 (K=6) + correction -> +6 tokens (161.5 ms) cpos=255
764
+ [margins] 16.31 16.87 12.54 8.25 3.74 K=6
765
+ verify 42: accepted 5/5 (K=6) + correction -> +6 tokens (183.1 ms) cpos=261
766
+ [margins] 16.66 10.03 8.10 8.21 0.58 K=5
767
+ verify 43: accepted 4/4 (K=5) + correction -> +5 tokens (146.0 ms) cpos=266
768
+ [margins] 12.32 9.33 2.69 5.28 6.15 K=6
769
+ verify 44: accepted 2/5 (K=6) + correction -> +3 tokens (168.2 ms) cpos=269
770
+ [margins] 15.17 11.97 4.16 9.50 10.34 K=6
771
+ verify 45: accepted 5/5 (K=6) + correction -> +6 tokens (165.5 ms) cpos=275
772
+ [margins] 13.10 18.46 13.98 9.48 4.77 K=6
773
+ verify 46: accepted 5/5 (K=6) + correction -> +6 tokens (169.3 ms) cpos=281
774
+ [margins] 16.63 7.05 7.50 7.95 0.91 K=5
775
+ verify 47: accepted 4/4 (K=5) + correction -> +5 tokens (145.6 ms) cpos=286
776
+
777
+ [spec] generated 204 tokens over 47 verifies: mean tokens/verify = 4.34 (block=5, max 6)
778
+ [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
779
+ [spec] SPEC-DECODE: 34.3 ms/tok = 29.18 tok/s (vs base M=1 68.6 ms/tok = 14.57 tok/s -> 2.00x)
780
+ [decode] PREFILL: 14 positions in 285.8 ms = 49.0 tok/s (20.414 ms/tok)
781
+ [spec] decoding 200 tokens (block=5, draft passes=1, adaptK=1.50, prompt=6 s=15)...
782
+ [margins] 0.32 7.93 3.07 0.82 5.18 K=4
783
+ verify 1: accepted 0/3 (K=4) + correction -> +1 tokens (123.9 ms) cpos=15
784
+ [margins] 1.44 0.38 2.32 0.28 0.76 K=2
785
+ verify 2: accepted 0/1 (K=2) + correction -> +1 tokens (104.1 ms) cpos=16
786
+ [margins] 3.72 0.25 0.77 0.64 1.05 K=2
787
+ verify 3: accepted 1/1 (K=2) + correction -> +2 tokens (98.9 ms) cpos=18
788
+ [margins] 2.29 0.28 0.22 1.28 1.74 K=2
789
+ verify 4: accepted 1/1 (K=2) + correction -> +2 tokens (103.0 ms) cpos=20
790
+ [margins] 3.78 0.29 2.08 0.82 2.33 K=2
791
+ verify 5: accepted 1/1 (K=2) + correction -> +2 tokens (98.8 ms) cpos=22
792
+ [margins] 3.48 1.43 1.82 1.46 3.01 K=2
793
+ verify 6: accepted 1/1 (K=2) + correction -> +2 tokens (102.1 ms) cpos=24
794
+ [margins] 3.90 4.08 2.28 0.26 2.22 K=4
795
+ verify 7: accepted 3/3 (K=4) + correction -> +4 tokens (117.7 ms) cpos=28
796
+ [margins] 3.28 6.35 3.13 3.72 2.35 K=6
797
+ verify 8: accepted 2/5 (K=6) + correction -> +3 tokens (144.1 ms) cpos=31
798
+ [margins] 2.48 0.97 3.63 2.35 2.08 K=2
799
+ verify 9: accepted 0/1 (K=2) + correction -> +1 tokens (99.0 ms) cpos=32
800
+ [margins] 8.12 0.51 1.45 1.10 0.14 K=2
801
+ verify 10: accepted 1/1 (K=2) + correction -> +2 tokens (96.9 ms) cpos=34
802
+ [margins] 0.25 6.09 1.43 0.52 1.01 K=3
803
+ verify 11: accepted 2/2 (K=3) + correction -> +3 tokens (111.1 ms) cpos=37
804
+ [margins] 11.66 6.73 1.60 2.63 0.60 K=5
805
+ verify 12: accepted 3/4 (K=5) + correction -> +4 tokens (133.3 ms) cpos=41
806
+ [margins] 9.43 7.72 10.88 7.63 6.46 K=6
807
+ verify 13: accepted 5/5 (K=6) + correction -> +6 tokens (150.3 ms) cpos=47
808
+ [margins] 10.49 10.00 10.50 9.28 7.40 K=6
809
+ verify 14: accepted 5/5 (K=6) + correction -> +6 tokens (157.0 ms) cpos=53
810
+ [margins] 10.71 10.69 8.15 10.36 7.29 K=6
811
+ verify 15: accepted 5/5 (K=6) + correction -> +6 tokens (149.5 ms) cpos=59
812
+ [margins] 11.46 10.64 11.12 8.21 6.39 K=6
813
+ verify 16: accepted 5/5 (K=6) + correction -> +6 tokens (157.4 ms) cpos=65
814
+ [margins] 9.71 10.86 9.03 7.93 6.90 K=6
815
+ verify 17: accepted 5/5 (K=6) + correction -> +6 tokens (155.1 ms) cpos=71
816
+ [margins] 11.26 12.10 13.05 8.57 5.44 K=6
817
+ verify 18: accepted 5/5 (K=6) + correction -> +6 tokens (160.9 ms) cpos=77
818
+ [margins] 10.01 10.91 11.18 9.21 5.11 K=6
819
+ verify 19: accepted 5/5 (K=6) + correction -> +6 tokens (152.6 ms) cpos=83
820
+ [margins] 10.46 12.82 9.93 9.53 3.89 K=6
821
+ verify 20: accepted 5/5 (K=6) + correction -> +6 tokens (163.1 ms) cpos=89
822
+ [margins] 11.69 12.76 13.76 9.67 7.11 K=6
823
+ verify 21: accepted 5/5 (K=6) + correction -> +6 tokens (155.1 ms) cpos=95
824
+ [margins] 11.60 11.46 13.53 10.75 8.18 K=6
825
+ verify 22: accepted 5/5 (K=6) + correction -> +6 tokens (162.3 ms) cpos=101
826
+ [margins] 9.47 11.90 10.98 8.23 4.96 K=6
827
+ verify 23: accepted 5/5 (K=6) + correction -> +6 tokens (155.4 ms) cpos=107
828
+ [margins] 11.48 12.65 12.07 10.30 6.88 K=6
829
+ verify 24: accepted 5/5 (K=6) + correction -> +6 tokens (163.2 ms) cpos=113
830
+ [margins] 10.59 9.97 14.86 9.32 7.23 K=6
831
+ verify 25: accepted 5/5 (K=6) + correction -> +6 tokens (154.5 ms) cpos=119
832
+ [margins] 10.54 11.38 9.45 9.76 6.88 K=6
833
+ verify 26: accepted 5/5 (K=6) + correction -> +6 tokens (163.3 ms) cpos=125
834
+ [margins] 11.80 11.53 12.53 9.30 8.42 K=6
835
+ verify 27: accepted 5/5 (K=6) + correction -> +6 tokens (170.1 ms) cpos=131
836
+ [margins] 11.39 11.84 11.89 9.26 5.51 K=6
837
+ verify 28: accepted 5/5 (K=6) + correction -> +6 tokens (162.0 ms) cpos=137
838
+ [margins] 11.72 13.61 12.48 9.89 8.95 K=6
839
+ verify 29: accepted 5/5 (K=6) + correction -> +6 tokens (154.6 ms) cpos=143
840
+ [margins] 11.95 12.37 13.28 10.28 8.48 K=6
841
+ verify 30: accepted 5/5 (K=6) + correction -> +6 tokens (161.9 ms) cpos=149
842
+ [margins] 10.92 12.46 12.51 9.57 6.06 K=6
843
+ verify 31: accepted 5/5 (K=6) + correction -> +6 tokens (154.8 ms) cpos=155
844
+ [margins] 11.85 11.98 13.41 8.58 7.51 K=6
845
+ verify 32: accepted 5/5 (K=6) + correction -> +6 tokens (164.7 ms) cpos=161
846
+ [margins] 10.80 12.44 12.97 10.15 8.00 K=6
847
+ verify 33: accepted 5/5 (K=6) + correction -> +6 tokens (157.2 ms) cpos=167
848
+ [margins] 11.51 12.97 10.32 9.25 8.34 K=6
849
+ verify 34: accepted 5/5 (K=6) + correction -> +6 tokens (163.6 ms) cpos=173
850
+ [margins] 11.13 13.54 12.46 8.45 3.95 K=6
851
+ verify 35: accepted 5/5 (K=6) + correction -> +6 tokens (157.3 ms) cpos=179
852
+ [margins] 11.88 12.29 12.80 10.32 4.99 K=6
853
+ verify 36: accepted 5/5 (K=6) + correction -> +6 tokens (164.4 ms) cpos=185
854
+ [margins] 12.29 12.14 13.36 9.73 7.75 K=6
855
+ verify 37: accepted 5/5 (K=6) + correction -> +6 tokens (160.0 ms) cpos=191
856
+ [margins] 10.79 12.82 11.86 8.57 5.53 K=6
857
+ verify 38: accepted 5/5 (K=6) + correction -> +6 tokens (164.5 ms) cpos=197
858
+ [margins] 10.38 11.56 11.37 9.49 6.13 K=6
859
+ verify 39: accepted 5/5 (K=6) + correction -> +6 tokens (158.3 ms) cpos=203
860
+ [margins] 11.67 12.52 10.93 8.83 5.50 K=6
861
+ verify 40: accepted 5/5 (K=6) + correction -> +6 tokens (164.3 ms) cpos=209
862
+ [margins] 10.85 12.81 13.39 8.69 3.37 K=6
863
+ verify 41: accepted 5/5 (K=6) + correction -> +6 tokens (158.9 ms) cpos=215
864
+
865
+ [spec] generated 201 tokens over 41 verifies: mean tokens/verify = 4.90 (block=5, max 6)
866
+ [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
867
+ [spec] SPEC-DECODE: 29.1 ms/tok = 34.33 tok/s (vs base M=1 68.6 ms/tok = 14.57 tok/s -> 2.36x)
868
+ [decode] PREFILL: 37 positions in 634.2 ms = 58.3 tok/s (17.140 ms/tok)
869
+ [spec] decoding 200 tokens (block=5, draft passes=1, adaptK=1.50, prompt=7 s=38)...
870
+ [margins] 0.14 0.24 0.43 0.01 0.45 K=2
871
+ verify 1: accepted 0/1 (K=2) + correction -> +1 tokens (100.4 ms) cpos=38
872
+ [margins] 0.77 6.09 0.34 1.09 0.34 K=3
873
+ verify 2: accepted 0/2 (K=3) + correction -> +1 tokens (111.4 ms) cpos=39
874
+ [margins] 1.40 0.28 5.25 0.37 0.62 K=2
875
+ verify 3: accepted 1/1 (K=2) + correction -> +2 tokens (99.8 ms) cpos=41
876
+ [margins] 3.42 0.18 0.21 6.65 0.01 K=2
877
+ verify 4: accepted 1/1 (K=2) + correction -> +2 tokens (96.1 ms) cpos=43
878
+ [margins] 1.07 1.59 0.53 0.24 0.19 K=3
879
+ verify 5: accepted 0/2 (K=3) + correction -> +1 tokens (111.4 ms) cpos=44
880
+ [margins] 0.36 0.01 0.82 0.15 0.24 K=2
881
+ verify 6: accepted 1/1 (K=2) + correction -> +2 tokens (95.6 ms) cpos=46
882
+ [margins] 1.11 1.39 1.56 0.27 0.66 K=2
883
+ verify 7: accepted 0/1 (K=2) + correction -> +1 tokens (101.9 ms) cpos=47
884
+ [margins] 3.40 2.61 0.25 0.45 1.72 K=3
885
+ verify 8: accepted 1/2 (K=3) + correction -> +2 tokens (113.9 ms) cpos=49
886
+ [margins] 8.08 7.02 4.41 2.04 0.87 K=5
887
+ verify 9: accepted 4/4 (K=5) + correction -> +5 tokens (138.3 ms) cpos=54
888
+ [margins] 0.76 0.29 1.30 2.37 0.79 K=2
889
+ verify 10: accepted 1/1 (K=2) + correction -> +2 tokens (102.4 ms) cpos=56
890
+ [margins] 5.39 2.63 1.15 0.38 0.52 K=3
891
+ verify 11: accepted 2/2 (K=3) + correction -> +3 tokens (107.8 ms) cpos=59
892
+ [margins] 3.99 2.31 0.09 0.21 0.06 K=3
893
+ verify 12: accepted 2/2 (K=3) + correction -> +3 tokens (113.6 ms) cpos=62
894
+ [margins] 2.27 2.17 4.15 0.97 2.55 K=4
895
+ verify 13: accepted 3/3 (K=4) + correction -> +4 tokens (126.6 ms) cpos=66
896
+ [margins] 13.86 3.56 5.65 0.54 0.70 K=4
897
+ verify 14: accepted 3/3 (K=4) + correction -> +4 tokens (129.1 ms) cpos=70
898
+ [margins] 0.82 0.82 0.12 1.56 0.44 K=2
899
+ verify 15: accepted 1/1 (K=2) + correction -> +2 tokens (101.9 ms) cpos=72
900
+ [margins] 2.44 6.11 0.22 0.23 0.15 K=3
901
+ verify 16: accepted 2/2 (K=3) + correction -> +3 tokens (109.0 ms) cpos=75
902
+ [margins] 1.86 8.25 0.09 1.18 1.64 K=3
903
+ verify 17: accepted 2/2 (K=3) + correction -> +3 tokens (112.1 ms) cpos=78
904
+ [margins] 7.75 7.75 5.59 8.01 8.69 K=6
905
+ verify 18: accepted 5/5 (K=6) + correction -> +6 tokens (160.8 ms) cpos=84
906
+ [margins] 9.07 8.06 6.36 5.57 0.09 K=5
907
+ verify 19: accepted 4/4 (K=5) + correction -> +5 tokens (142.3 ms) cpos=89
908
+ [margins] 8.42 5.82 4.97 0.82 0.80 K=4
909
+ verify 20: accepted 3/3 (K=4) + correction -> +4 tokens (128.6 ms) cpos=93
910
+ [margins] 4.49 3.09 0.94 0.41 0.79 K=3
911
+ verify 21: accepted 2/2 (K=3) + correction -> +3 tokens (115.2 ms) cpos=96
912
+ [margins] 10.55 8.36 5.92 7.71 10.32 K=6
913
+ verify 22: accepted 5/5 (K=6) + correction -> +6 tokens (160.2 ms) cpos=102
914
+ [margins] 16.78 9.32 4.05 2.78 2.30 K=6
915
+ verify 23: accepted 5/5 (K=6) + correction -> +6 tokens (165.8 ms) cpos=108
916
+ [margins] 9.52 10.75 3.40 2.44 3.91 K=6
917
+ verify 24: accepted 5/5 (K=6) + correction -> +6 tokens (161.9 ms) cpos=114
918
+ [margins] 12.05 4.17 5.41 6.23 9.15 K=6
919
+ verify 25: accepted 5/5 (K=6) + correction -> +6 tokens (165.5 ms) cpos=120
920
+ [margins] 14.58 9.44 6.89 9.43 0.29 K=5
921
+ verify 26: accepted 4/4 (K=5) + correction -> +5 tokens (144.1 ms) cpos=125
922
+ [margins] 6.91 7.95 2.31 6.31 3.64 K=6
923
+ verify 27: accepted 5/5 (K=6) + correction -> +6 tokens (175.0 ms) cpos=131
924
+ [margins] 0.35 6.30 0.62 3.08 1.19 K=3
925
+ verify 28: accepted 2/2 (K=3) + correction -> +3 tokens (116.3 ms) cpos=134
926
+ [margins] 13.18 5.27 0.58 0.64 1.17 K=3
927
+ verify 29: accepted 2/2 (K=3) + correction -> +3 tokens (119.0 ms) cpos=137
928
+ [margins] 2.56 0.15 1.01 0.21 0.64 K=2
929
+ verify 30: accepted 1/1 (K=2) + correction -> +2 tokens (103.0 ms) cpos=139
930
+ [margins] 7.66 0.50 6.12 7.11 4.79 K=2
931
+ verify 31: accepted 1/1 (K=2) + correction -> +2 tokens (104.3 ms) cpos=141
932
+ [margins] 10.01 8.87 5.03 2.06 0.80 K=5
933
+ verify 32: accepted 4/4 (K=5) + correction -> +5 tokens (146.0 ms) cpos=146
934
+ [margins] 5.98 10.82 10.20 11.03 2.76 K=6
935
+ verify 33: accepted 4/5 (K=6) + correction -> +5 tokens (169.2 ms) cpos=151
936
+ [margins] 2.40 0.03 0.06 0.41 1.88 K=2
937
+ verify 34: accepted 1/1 (K=2) + correction -> +2 tokens (110.0 ms) cpos=153
938
+ [margins] 2.50 4.99 2.40 0.05 1.07 K=4
939
+ verify 35: accepted 0/3 (K=4) + correction -> +1 tokens (120.6 ms) cpos=154
940
+ [margins] 7.38 2.46 0.28 1.52 2.06 K=3
941
+ verify 36: accepted 1/2 (K=3) + correction -> +2 tokens (115.6 ms) cpos=156
942
+ [margins] 6.30 4.43 0.57 0.07 3.04 K=3
943
+ verify 37: accepted 1/2 (K=3) + correction -> +2 tokens (109.9 ms) cpos=158
944
+ [margins] 4.39 0.32 3.44 0.28 1.69 K=2
945
+ verify 38: accepted 1/1 (K=2) + correction -> +2 tokens (108.2 ms) cpos=160
946
+ [margins] 5.00 4.43 3.60 1.31 2.83 K=4
947
+ verify 39: accepted 3/3 (K=4) + correction -> +4 tokens (124.7 ms) cpos=164
948
+ [margins] 3.48 7.07 4.15 4.40 3.93 K=6
949
+ verify 40: accepted 5/5 (K=6) + correction -> +6 tokens (149.4 ms) cpos=170
950
+ [margins] 2.28 8.11 1.80 6.74 4.28 K=6
951
+ verify 41: accepted 5/5 (K=6) + correction -> +6 tokens (155.7 ms) cpos=176
952
+ [margins] 2.12 9.33 2.62 6.35 3.90 K=6
953
+ verify 42: accepted 5/5 (K=6) + correction -> +6 tokens (150.2 ms) cpos=182
954
+ [margins] 3.16 9.65 3.45 6.38 4.73 K=6
955
+ verify 43: accepted 5/5 (K=6) + correction -> +6 tokens (156.5 ms) cpos=188
956
+ [margins] 3.12 8.81 3.30 5.90 5.11 K=6
957
+ verify 44: accepted 5/5 (K=6) + correction -> +6 tokens (150.1 ms) cpos=194
958
+ [margins] 2.99 9.57 2.92 6.75 4.69 K=6
959
+ verify 45: accepted 5/5 (K=6) + correction -> +6 tokens (156.2 ms) cpos=200
960
+ [margins] 2.48 9.20 3.18 6.80 4.17 K=6
961
+ verify 46: accepted 5/5 (K=6) + correction -> +6 tokens (150.5 ms) cpos=206
962
+ [margins] 2.40 9.47 2.82 7.30 4.35 K=6
963
+ verify 47: accepted 5/5 (K=6) + correction -> +6 tokens (157.1 ms) cpos=212
964
+ [margins] 2.79 9.74 2.77 7.22 4.29 K=6
965
+ verify 48: accepted 5/5 (K=6) + correction -> +6 tokens (149.4 ms) cpos=218
966
+ [margins] 2.78 9.73 4.21 6.44 3.98 K=6
967
+ verify 49: accepted 5/5 (K=6) + correction -> +6 tokens (156.4 ms) cpos=224
968
+ [margins] 2.58 10.59 2.85 6.99 3.44 K=6
969
+ verify 50: accepted 5/5 (K=6) + correction -> +6 tokens (150.1 ms) cpos=230
970
+ [margins] 3.13 9.87 3.44 6.19 3.83 K=6
971
+ verify 51: accepted 0/5 (K=6) + correction -> +1 tokens (156.7 ms) cpos=231
972
+ [margins] 8.66 1.83 3.78 4.26 4.17 K=6
973
+ verify 52: accepted 0/5 (K=6) + correction -> +1 tokens (161.9 ms) cpos=232
974
+ [margins] 6.69 0.60 8.18 1.47 1.34 K=2
975
+ verify 53: accepted 1/1 (K=2) + correction -> +2 tokens (101.2 ms) cpos=234
976
+ [margins] 5.78 0.99 3.39 0.80 1.14 K=2
977
+ verify 54: accepted 1/1 (K=2) + correction -> +2 tokens (105.6 ms) cpos=236
978
+ [margins] 4.58 0.19 8.22 0.30 0.87 K=2
979
+ verify 55: accepted 0/1 (K=2) + correction -> +1 tokens (101.4 ms) cpos=237
980
+
981
+ [spec] generated 200 tokens over 55 verifies: mean tokens/verify = 3.64 (block=5, max 6)
982
+ [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
983
+ [spec] SPEC-DECODE: 35.4 ms/tok = 28.24 tok/s (vs base M=1 68.6 ms/tok = 14.57 tok/s -> 1.94x)
984
+ [decode] PREFILL: 12 positions in 252.5 ms = 47.5 tok/s (21.044 ms/tok)
985
+ [spec] decoding 200 tokens (block=5, draft passes=1, adaptK=1.50, prompt=8 s=13)...
986
+ [margins] 2.65 0.36 0.13 0.93 1.09 K=2
987
+ verify 1: accepted 1/1 (K=2) + correction -> +2 tokens (99.0 ms) cpos=14
988
+ [margins] 0.65 6.84 3.26 0.54 5.69 K=4
989
+ verify 2: accepted 0/3 (K=4) + correction -> +1 tokens (127.8 ms) cpos=15
990
+ [margins] 2.65 2.80 0.02 2.28 1.27 K=3
991
+ verify 3: accepted 0/2 (K=3) + correction -> +1 tokens (103.4 ms) cpos=16
992
+ [margins] 3.81 1.15 0.56 0.50 0.89 K=2
993
+ verify 4: accepted 0/1 (K=2) + correction -> +1 tokens (95.2 ms) cpos=17
994
+ [margins] 4.04 4.39 0.17 3.01 0.47 K=3
995
+ verify 5: accepted 0/2 (K=3) + correction -> +1 tokens (102.7 ms) cpos=18
996
+ [margins] 0.14 0.04 0.17 0.07 0.99 K=2
997
+ verify 6: accepted 0/1 (K=2) + correction -> +1 tokens (99.6 ms) cpos=19
998
+ [margins] 0.01 3.21 0.61 0.09 0.47 K=3
999
+ verify 7: accepted 2/2 (K=3) + correction -> +3 tokens (108.7 ms) cpos=22
1000
+ [margins] 1.35 5.34 0.78 4.09 0.76 K=3
1001
+ verify 8: accepted 0/2 (K=3) + correction -> +1 tokens (108.9 ms) cpos=23
1002
+ [margins] 0.49 1.37 0.68 0.64 0.65 K=2
1003
+ verify 9: accepted 1/1 (K=2) + correction -> +2 tokens (103.0 ms) cpos=25
1004
+ [margins] 1.98 1.62 1.78 0.48 0.80 K=4
1005
+ verify 10: accepted 3/3 (K=4) + correction -> +4 tokens (118.1 ms) cpos=29
1006
+ [margins] 4.31 1.27 3.87 0.63 4.62 K=2
1007
+ verify 11: accepted 1/1 (K=2) + correction -> +2 tokens (98.9 ms) cpos=31
1008
+ [margins] 4.89 1.51 4.58 0.34 4.66 K=4
1009
+ verify 12: accepted 3/3 (K=4) + correction -> +4 tokens (118.8 ms) cpos=35
1010
+ [margins] 6.25 1.08 5.86 0.42 5.35 K=2
1011
+ verify 13: accepted 1/1 (K=2) + correction -> +2 tokens (102.9 ms) cpos=37
1012
+ [margins] 5.69 1.95 5.05 0.58 4.52 K=4
1013
+ verify 14: accepted 3/3 (K=4) + correction -> +4 tokens (118.8 ms) cpos=41
1014
+ [margins] 5.79 1.54 5.18 0.83 4.48 K=4
1015
+ verify 15: accepted 3/3 (K=4) + correction -> +4 tokens (121.9 ms) cpos=45
1016
+ [margins] 6.84 0.72 4.98 1.62 3.96 K=2
1017
+ verify 16: accepted 0/1 (K=2) + correction -> +1 tokens (100.3 ms) cpos=46
1018
+ [margins] 5.25 3.76 6.07 1.10 1.25 K=4
1019
+ verify 17: accepted 3/3 (K=4) + correction -> +4 tokens (128.8 ms) cpos=50
1020
+ [margins] 2.95 1.73 0.69 0.65 1.73 K=3
1021
+ verify 18: accepted 2/2 (K=3) + correction -> +3 tokens (113.6 ms) cpos=53
1022
+ [margins] 8.07 9.22 0.06 3.51 0.49 K=3
1023
+ verify 19: accepted 2/2 (K=3) + correction -> +3 tokens (111.8 ms) cpos=56
1024
+ [margins] 3.30 1.47 2.98 0.50 1.76 K=2
1025
+ verify 20: accepted 1/1 (K=2) + correction -> +2 tokens (100.5 ms) cpos=58
1026
+ [margins] 0.19 7.64 0.11 0.43 1.07 K=3
1027
+ verify 21: accepted 0/2 (K=3) + correction -> +1 tokens (111.0 ms) cpos=59
1028
+ [margins] 1.69 1.49 0.93 1.39 0.38 K=2
1029
+ verify 22: accepted 0/1 (K=2) + correction -> +1 tokens (102.1 ms) cpos=60
1030
+ [margins] 4.86 1.59 6.61 7.44 3.82 K=6
1031
+ verify 23: accepted 5/5 (K=6) + correction -> +6 tokens (154.6 ms) cpos=66
1032
+ [margins] 8.24 9.58 9.47 1.67 0.06 K=5
1033
+ verify 24: accepted 3/4 (K=5) + correction -> +4 tokens (141.3 ms) cpos=70
1034
+ [margins] 0.65 5.63 0.42 3.44 0.34 K=3
1035
+ verify 25: accepted 0/2 (K=3) + correction -> +1 tokens (115.3 ms) cpos=71
1036
+ [margins] 1.52 4.28 4.21 2.24 2.26 K=6
1037
+ verify 26: accepted 0/5 (K=6) + correction -> +1 tokens (161.8 ms) cpos=72
1038
+ [margins] 3.36 1.66 5.24 7.42 3.05 K=6
1039
+ verify 27: accepted 5/5 (K=6) + correction -> +6 tokens (156.5 ms) cpos=78
1040
+ [margins] 7.11 9.29 7.41 0.96 0.01 K=4
1041
+ verify 28: accepted 3/3 (K=4) + correction -> +4 tokens (129.2 ms) cpos=82
1042
+ [margins] 3.31 5.41 1.43 3.42 3.61 K=3
1043
+ verify 29: accepted 2/2 (K=3) + correction -> +3 tokens (115.5 ms) cpos=85
1044
+ [margins] 4.40 8.18 1.86 2.50 6.74 K=6
1045
+ verify 30: accepted 5/5 (K=6) + correction -> +6 tokens (157.1 ms) cpos=91
1046
+ [margins] 4.32 12.56 0.85 2.80 4.24 K=3
1047
+ verify 31: accepted 2/2 (K=3) + correction -> +3 tokens (116.5 ms) cpos=94
1048
+ [margins] 4.75 6.79 7.00 4.04 3.57 K=6
1049
+ verify 32: accepted 5/5 (K=6) + correction -> +6 tokens (165.2 ms) cpos=100
1050
+ [margins] 4.95 5.95 3.53 5.86 0.98 K=5
1051
+ verify 33: accepted 4/4 (K=5) + correction -> +5 tokens (143.8 ms) cpos=105
1052
+ [margins] 3.32 4.96 3.67 4.19 6.55 K=6
1053
+ verify 34: accepted 5/5 (K=6) + correction -> +6 tokens (159.4 ms) cpos=111
1054
+ [margins] 8.04 2.58 3.61 0.31 4.82 K=4
1055
+ verify 35: accepted 3/3 (K=4) + correction -> +4 tokens (131.7 ms) cpos=115
1056
+ [margins] 5.58 15.40 2.18 5.38 8.24 K=6
1057
+ verify 36: accepted 5/5 (K=6) + correction -> +6 tokens (166.4 ms) cpos=121
1058
+ [margins] 5.38 9.72 5.78 5.80 6.19 K=6
1059
+ verify 37: accepted 5/5 (K=6) + correction -> +6 tokens (158.8 ms) cpos=127
1060
+ [margins] 5.38 14.53 2.83 6.26 9.39 K=6
1061
+ verify 38: accepted 5/5 (K=6) + correction -> +6 tokens (182.9 ms) cpos=133
1062
+ [margins] 5.56 11.98 7.60 4.13 7.13 K=6
1063
+ verify 39: accepted 5/5 (K=6) + correction -> +6 tokens (160.9 ms) cpos=139
1064
+ [margins] 5.51 9.71 2.57 5.85 7.31 K=6
1065
+ verify 40: accepted 5/5 (K=6) + correction -> +6 tokens (167.6 ms) cpos=145
1066
+ [margins] 7.35 10.60 7.46 3.50 8.42 K=6
1067
+ verify 41: accepted 5/5 (K=6) + correction -> +6 tokens (160.5 ms) cpos=151
1068
+ [margins] 5.69 13.35 2.07 7.00 6.68 K=6
1069
+ verify 42: accepted 5/5 (K=6) + correction -> +6 tokens (167.1 ms) cpos=157
1070
+ [margins] 8.78 10.47 6.80 4.53 7.31 K=6
1071
+ verify 43: accepted 5/5 (K=6) + correction -> +6 tokens (162.5 ms) cpos=163
1072
+ [margins] 5.80 12.78 2.58 7.14 6.30 K=6
1073
+ verify 44: accepted 5/5 (K=6) + correction -> +6 tokens (167.6 ms) cpos=169
1074
+ [margins] 8.93 8.05 8.46 4.72 1.24 K=5
1075
+ verify 45: accepted 4/4 (K=5) + correction -> +5 tokens (144.0 ms) cpos=174
1076
+ [margins] 13.94 12.08 9.86 1.62 5.07 K=6
1077
+ verify 46: accepted 5/5 (K=6) + correction -> +6 tokens (166.9 ms) cpos=180
1078
+ [margins] 5.89 8.82 10.16 8.93 5.65 K=6
1079
+ verify 47: accepted 5/5 (K=6) + correction -> +6 tokens (161.6 ms) cpos=186
1080
+ [margins] 12.86 9.29 9.80 1.98 5.70 K=6
1081
+ verify 48: accepted 5/5 (K=6) + correction -> +6 tokens (167.1 ms) cpos=192
1082
+ [margins] 4.69 7.94 9.50 8.28 6.24 K=6
1083
+ verify 49: accepted 5/5 (K=6) + correction -> +6 tokens (162.2 ms) cpos=198
1084
+ [margins] 15.25 10.40 8.74 2.02 5.14 K=6
1085
+ verify 50: accepted 5/5 (K=6) + correction -> +6 tokens (166.8 ms) cpos=204
1086
+ [margins] 4.29 8.18 9.53 8.22 6.69 K=6
1087
+ verify 51: accepted 5/5 (K=6) + correction -> +6 tokens (163.3 ms) cpos=210
1088
+ [margins] 13.47 11.18 8.65 2.02 4.80 K=6
1089
+ verify 52: accepted 5/5 (K=6) + correction -> +6 tokens (167.7 ms) cpos=216
1090
+
1091
+ [spec] generated 204 tokens over 52 verifies: mean tokens/verify = 3.92 (block=5, max 6)
1092
+ [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
1093
+ [spec] SPEC-DECODE: 34.3 ms/tok = 29.11 tok/s (vs base M=1 68.6 ms/tok = 14.57 tok/s -> 2.00x)
1094
+
1095
+ [blksweep] BLK passes adaptK->eff prompt | mean tok/verify | ms/tok | tok/s | vs base 14.57 tok/s
1096
+ [blksweep] 5 1 1.50 1.50 0 | 3.26 | 39.5 | 25.35 | 1.74x
1097
+ [blksweep] 5 1 1.50 1.50 1 | 4.00 | 34.8 | 28.72 | 1.97x
1098
+ [blksweep] 5 1 1.50 1.50 2 | 4.04 | 35.1 | 28.45 | 1.95x
1099
+ [blksweep] 5 1 1.50 1.50 3 | 2.52 | 47.4 | 21.08 | 1.45x
1100
+ [blksweep] 5 1 1.50 1.50 4 | 3.06 | 39.9 | 25.04 | 1.72x
1101
+ [blksweep] 5 1 1.50 1.50 5 | 4.34 | 34.3 | 29.18 | 2.00x
1102
+ [blksweep] 5 1 1.50 1.50 6 | 4.90 | 29.1 | 34.33 | 2.36x
1103
+ [blksweep] 5 1 1.50 1.50 7 | 3.64 | 35.4 | 28.24 | 1.94x
1104
+ [blksweep] 5 1 1.50 1.50 8 | 3.92 | 34.3 | 29.11 | 2.00x
1105
+ [decode] mem 106.6/122.8 GiB