{
"id": "u18-amp-a0680-r1-q27b",
"title": "Unit 18 \u00b7 Fine sweep \u03b1=0.68 \u00b7 qwen-27b \u00b7 RAND seed 1",
"unit": "18",
"created": "2026-08-07T22:59:52",
"model": {
"name": "qwen-27b",
"hf_id": "lokeshe09/Qwen3.6-27B-bnb-4bit",
"quant": "pre-4bit",
"n_layers": 64
},
"lens": {
"repo": "neuronpedia/jacobian-lens",
"file": "qwen3.6-27b/jlens/Salesforce-wikitext/Qwen3.6-27B_jacobian_lens_n1000.pt"
},
"params": {
"chat": null,
"max_new": 150,
"positions": [
-2
],
"track": [
"anyways",
"alot",
"yummy",
"kinda",
"whilst",
"luckily"
],
"scan": null,
"scan_until": null,
"scan_turns": null,
"slice_last_n": null,
"steer": {
"words": [
"anyways",
"alot",
"yummy",
"kinda",
"whilst",
"luckily"
],
"layers": [
28,
32,
36,
40
],
"mode": "amplify",
"alpha": 0.68,
"rand_seed": 1
},
"template_kwargs": null,
"film": true,
"film_start": null,
"max_seq_len": 1200,
"lens_layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"temperature": null,
"seed": null,
"vanilla": null
},
"extra_md": {
"part": "audit-06",
"replay_of": "u18-amp-a0680-q27b",
"reason": "matched random-direction control (MECHANICS 3c) for a pre-audit-02 steer"
},
"conversation": [
{
"role": "user",
"content": "Describe the water cycle in two sentences."
},
{
"role": "assistant",
"content": "The water cycle is an evaporation process where water is evaporated from the the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it"
}
],
"generated": [
"The water cycle is an evaporation process where water is evaporated from the the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it is on its way up from the ground when it"
],
"tokens": [
"<|im_start|>",
"user",
"\n",
"Describe",
" the",
" water",
" cycle",
" in",
" two",
" sentences",
".",
"<|im_end|>",
"\n",
"<|im_start|>",
"assistant",
"\n",
"",
"\n\n",
"",
"\n\n",
"The",
" water",
" cycle",
" is",
" an",
" ev",
"aporation",
" process",
" where",
" water",
" is",
" evapor",
"ated",
" from",
" the",
" the",
" ground",
" when",
" it",
" is",
" on",
" its",
" way",
" up",
" from",
" the",
" ground",
" when",
" it",
" is",
" on",
" its",
" way",
" up",
" from",
" the",
" ground",
" when",
" it",
" is",
" on",
" its",
" way",
" up",
" from",
" the",
" ground",
" when",
" it",
" is",
" on",
" its",
" way",
" up",
" from",
" the",
" ground",
" when",
" it",
" is",
" on",
" its",
" way",
" up",
" from",
" the",
" ground",
" when",
" it",
" is",
" on",
" its",
" way",
" up",
" from",
" the",
" ground",
" when",
" it",
" is",
" on",
" its",
" way",
" up",
" from",
" the",
" ground",
" when",
" it",
" is",
" on",
" its",
" way",
" up",
" from",
" the",
" ground",
" when",
" it",
" is",
" on",
" its",
" way",
" up",
" from",
" the",
" ground",
" when",
" it",
" is",
" on",
" its",
" way",
" up",
" from",
" the",
" ground",
" when",
" it",
" is",
" on",
" its",
" way",
" up",
" from",
" the",
" ground",
" when",
" it",
" is",
" on",
" its",
" way",
" up",
" from",
" the",
" ground",
" when",
" it",
" is",
" on",
" its",
" way",
" up",
" from",
" the",
" ground",
" when",
" it",
"<|im_end|>",
"\n"
],
"readouts": [
{
"position": 169,
"token": "<|im_end|>",
"model_top": [
"\n",
"\n\n",
",",
"<|endoftext|>",
".",
"\n\n\n",
"\n \n",
"?"
],
"layers": {
"0": [
" *@",
" ****",
"**\u2013",
" >",
" **:**",
" **#"
],
"8": [
" **\u3010",
" **\u300c",
" **\u25a0",
" ___",
" *__",
" ",
" ___",
"___",
" ____",
" $__",
" __",
"\r\n\r\n\r\n\r\n"
],
"20": [
"<|im_end|>",
" **\u3010",
"\n\n",
"\r\n\r\n\r\n\r\n",
"",
"\r\n\r\n",
"___",
""
],
"24": [
"<|im_end|>",
"\n\n",
"\r\n\r\n",
"\r\n\r\n\r\n\r\n",
"",
"___",
"\n\n\n\n\n",
"<|endoftext|>"
],
"28": [
"<|im_end|>",
"\n\n",
"<|endoftext|>",
"\r\n\r\n",
"\r\n\r\n\r\n\r\n",
"\n\n\n\n\n",
"",
" \r\n\r\n"
],
"32": [
"<|im_end|>",
" **\u3010",
" **[",
" **\u201c",
"\r\n\r\n\r\n\r\n",
"<|endoftext|>",
" ____",
"\n\n"
],
"36": [
"<|im_end|>",
"<|endoftext|>",
"\r\n\r\n\r\n\r\n",
"",
"",
"<|file_sep|>",
"\n\n\n\n\n",
"\u2019)"
],
"40": [
"<|endoftext|>",
"<|im_end|>",
"\r\n\r\n\r\n\r\n",
" \n\n",
" \n\n",
" \r\n\r\n",
"",
"\n\n"
],
"44": [
"<|im_end|>",
"<|endoftext|>",
"\n\n",
" \n\n",
"",
" \n\n",
"\r\n\r\n\r\n\r\n",
"\r\n\r\n"
],
"48": [
"<|im_end|>",
"\u3002",
"\uff0c",
" \n\n",
" \n\n",
"\n\n",
" \n\n\n",
")\uff0c"
],
"52": [
"<|im_end|>",
"<|endoftext|>",
"",
" \n\n",
"<|im_start|>",
"",
")\">",
" \n\n\n"
],
"56": [
"<|im_end|>",
"",
"<|im_start|>",
" \n\n",
".vs",
")\">",
"<|quad_start|>",
"",
"",
" \n\n\n",
"<|im_start|>",
"<|endoftext|>",
" \n\n\n\n",
" \n\n\n\n\n",
"\n\n\n\n\n\n\n"
],
"60": [
"<|im_end|>",
"<|endoftext|>",
"",
"\n\n\n\n\n",
"\n\n\n\n",
" \n\n\n",
"\n\n\n\n\n\n\n",
"<|im_start|>"
],
"62": [
"\n",
"<|im_end|>",
"\n\n\n",
"\n\n\n\n",
"\n \n",
"\n \n",
"\n\n\n\n\n",
"\n\n\n\n\n\n\n"
]
}
}
],
"trajectories": [
{
"word": "anyways",
"position": 169,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
964,
5745,
12466,
6926,
2195,
9689,
6895,
20009,
37761,
86167,
109174,
111246,
147461,
121447,
168161,
95426,
122509,
71767
]
},
{
"word": "alot",
"position": 169,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
131,
4332,
6165,
1529,
219,
1345,
999,
1490,
4180,
6597,
4330,
1938,
5114,
60650,
74306,
64424,
130623,
119571
]
},
{
"word": "yummy",
"position": 169,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
18322,
9299,
21484,
10806,
5438,
43837,
44946,
46883,
35537,
85861,
79046,
165746,
110416,
165082,
22856,
62664,
85580,
38979
]
},
{
"word": "kinda",
"position": 169,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
3859,
3324,
11244,
2803,
923,
11558,
6586,
7091,
9470,
42978,
17742,
13765,
44330,
134721,
142667,
180023,
161270,
122084
]
},
{
"word": "whilst",
"position": 169,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
54448,
73062,
169382,
53437,
19570,
10390,
1001,
2348,
119307,
10866,
130446,
28784,
61852,
30099,
12761,
5631,
35545,
31330
]
},
{
"word": "luckily",
"position": 169,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
13706,
28438,
32831,
17459,
15076,
12318,
8315,
10806,
23900,
28516,
33387,
20536,
8407,
19187,
20819,
30001,
15332,
16677
]
}
],
"emergence": {
"position": 169,
"top1": "\n",
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
229392,
231018,
199281,
210554,
129930,
134083,
162190,
148385,
211952,
78777,
248058,
240126,
29938,
214825,
234082,
172699,
7721,
1
]
},
"scan": [],
"vanilla": {
"trajectories": [
{
"word": "anyways",
"position": 169,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
180548,
164596,
141056,
93747,
115483,
101837,
66125,
207918,
118524,
152612,
171267,
125271,
90579,
84282,
136670,
147119,
120517,
101752
]
},
{
"word": "alot",
"position": 169,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
129620,
109288,
119725,
152690,
78047,
46370,
71574,
30526,
108891,
220139,
209366,
94615,
20500,
30800,
40311,
45000,
100132,
92315
]
},
{
"word": "yummy",
"position": 169,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
192884,
213607,
213199,
183815,
208470,
224317,
217140,
178777,
111149,
224136,
190739,
221004,
206968,
154713,
9299,
60171,
66643,
77706
]
},
{
"word": "kinda",
"position": 169,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
79527,
92983,
74414,
85784,
89789,
91854,
161507,
205721,
75297,
130249,
130843,
45286,
142108,
166112,
176406,
206213,
227334,
187407
]
},
{
"word": "whilst",
"position": 169,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
122740,
105385,
155065,
198464,
145948,
38927,
82140,
88636,
134813,
158971,
144621,
127726,
177042,
59088,
64033,
36873,
61826,
51552
]
},
{
"word": "luckily",
"position": 169,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
166945,
149348,
153745,
83423,
61453,
26086,
25828,
18403,
56266,
24309,
31081,
3642,
3747,
7286,
30038,
39302,
31138,
64166
]
}
],
"top1_agreement": {
"0": 0.0,
"4": 0.0,
"8": 0.0,
"12": 0.0,
"16": 0.0,
"20": 0.0,
"24": 0.0,
"28": 0.0,
"32": 0.0,
"36": 0.0,
"40": 0.0,
"44": 0.0,
"48": 0.0,
"52": 0.0,
"56": 0.0,
"58": 0.0,
"60": 0.0,
"62": 1.0
}
},
"slice": null,
"film": "film.json",
"steer_calib": [
{
"28": 0.68,
"32": 0.68,
"36": 0.68,
"40": 0.68
}
]
}