{
"id": "u18-amp-a0454-r1-q27b",
"title": "Unit 18 \u00b7 Fine sweep \u03b1=0.4536 \u00b7 qwen-27b \u00b7 RAND seed 1",
"unit": "18",
"created": "2026-08-07T22:58:30",
"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": 0.4536,
"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-a0454-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 a continuous process where water evaporates from oceans and lakes, rises into the atmosphere as vapor, and then condenses into clouds. When precipitation falls, the water returns to the surface, completing the cycle as it flows back into rivers and oceans.\n\n**Note:** The prompt asks to describe the water cycle in two sentences. The provided text is a two-sentence description of the water cycle.\n\n**Revised Output (Two Sentences):**\n\nThe water cycle is a continuous process where water evaporates from oceans and lakes, rises into the atmosphere as vapor, and then condenses into clouds. When precipitation falls, the water returns to the surface, completing the cycle as it flows back into rivers and oceans.\n\n**Wait"
}
],
"generated": [
"The water cycle is a continuous process where water evaporates from oceans and lakes, rises into the atmosphere as vapor, and then condenses into clouds. When precipitation falls, the water returns to the surface, completing the cycle as it flows back into rivers and oceans.\n\n**Note:** The prompt asks to describe the water cycle in two sentences. The provided text is a two-sentence description of the water cycle.\n\n**Revised Output (Two Sentences):**\n\nThe water cycle is a continuous process where water evaporates from oceans and lakes, rises into the atmosphere as vapor, and then condenses into clouds. When precipitation falls, the water returns to the surface, completing the cycle as it flows back into rivers and oceans.\n\n**Wait"
],
"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",
" a",
" continuous",
" process",
" where",
" water",
" evapor",
"ates",
" from",
" oceans",
" and",
" lakes",
",",
" rises",
" into",
" the",
" atmosphere",
" as",
" vapor",
",",
" and",
" then",
" cond",
"enses",
" into",
" clouds",
".",
" When",
" precipitation",
" falls",
",",
" the",
" water",
" returns",
" to",
" the",
" surface",
",",
" completing",
" the",
" cycle",
" as",
" it",
" flows",
" back",
" into",
" rivers",
" and",
" oceans",
".",
"\n\n",
"**",
"Note",
":**",
" The",
" prompt",
" asks",
" to",
" describe",
" the",
" water",
" cycle",
" in",
" two",
" sentences",
".",
" The",
" provided",
" text",
" is",
" a",
" two",
"-s",
"entence",
" description",
" of",
" the",
" water",
" cycle",
".",
"\n\n",
"**",
"Re",
"vised",
" Output",
" (",
"Two",
" Sent",
"ences",
"):**",
"\n\n",
"The",
" water",
" cycle",
" is",
" a",
" continuous",
" process",
" where",
" water",
" evapor",
"ates",
" from",
" oceans",
" and",
" lakes",
",",
" rises",
" into",
" the",
" atmosphere",
" as",
" vapor",
",",
" and",
" then",
" cond",
"enses",
" into",
" clouds",
".",
" When",
" precipitation",
" falls",
",",
" the",
" water",
" returns",
" to",
" the",
" surface",
",",
" completing",
" the",
" cycle",
" as",
" it",
" flows",
" back",
" into",
" rivers",
" and",
" oceans",
".",
"\n\n",
"**",
"Wait",
"<|im_end|>",
"\n"
],
"readouts": [
{
"position": 170,
"token": "<|im_end|>",
"model_top": [
"\n",
"\n\n",
"<|endoftext|>",
" \n",
"\n \n",
".",
" ",
","
],
"layers": {
"0": [
" >",
"**\u201d,",
" **)",
" **:",
" /**<"
],
"8": [
" **\u3010",
" **\u300c",
" **\u25a0",
"(___",
"**\u201d,",
" **>>",
" **\u201c",
" ",
" **\u3010",
"___",
"",
"",
"\r\n\r\n\r\n\r\n",
"\uff1f**",
"**\u3002"
],
"20": [
"<|im_end|>",
"",
" **\u3010",
"\r\n\r\n\r\n\r\n",
"",
"___",
"\u548c",
"\u5728"
],
"24": [
"<|im_end|>",
"\r\n\r\n\r\n\r\n",
"___",
"",
"\r\n\r\n",
" \r\n\r\n",
"\n\n",
" \n\n"
],
"28": [
"<|im_end|>",
"\n\n",
"<|endoftext|>",
"\r\n\r\n\r\n\r\n",
"\r\n\r\n",
" \n\n",
"",
"___"
],
"32": [
"<|im_end|>",
"<|endoftext|>",
"\r\n\r\n\r\n\r\n",
"",
"\uff1f",
"\uff1f**",
" \n\n",
" \n\n\n"
],
"36": [
"<|im_end|>",
"<|endoftext|>",
"",
"\n\n",
" \n\n",
"\r\n\r\n\r\n\r\n",
" \n\n",
"<|im_start|>"
],
"40": [
"<|endoftext|>",
"<|im_end|>",
" \n\n",
" \n\n",
"",
"\r\n\r\n\r\n\r\n",
" \r\n\r\n",
"\n\n"
],
"44": [
"<|endoftext|>",
"<|im_end|>",
" \n\n",
"",
" \n\n",
"\n\n",
"\uff0c",
"\r\n\r\n\r\n\r\n"
],
"48": [
"<|im_end|>",
"\u3002",
"\uff0c",
")\uff0c",
"",
")\u3002",
"\u5728",
"\uff1f"
],
"52": [
"<|endoftext|>",
"<|im_end|>",
"",
"<|im_start|>",
"<|file_sep|>",
"\u3002",
"",
" \n\n"
],
"56": [
"<|im_end|>",
"",
"<|im_start|>",
"ording",
"<|quad_start|>",
"\u5e0c\u671b",
"_hop",
"\u5e0c\u671b\u7684"
],
"58": [
"<|im_end|>",
"",
"\u89e3\u8bfb",
"Explanation",
"ording",
"\u91ca",
"\u89e3\u91ca",
"<|endoftext|>"
],
"60": [
"<|endoftext|>",
"",
"<|im_end|>",
"-hide",
"<|im_start|>",
"ording",
"yip",
"udos"
],
"62": [
"\n",
" \n",
"<|im_end|>",
"\n \n",
"-hide",
"",
" \n",
" trailing"
]
}
}
],
"trajectories": [
{
"word": "anyways",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
1553,
4008,
7528,
8605,
3684,
12368,
5668,
12153,
10917,
27460,
57788,
121856,
144919,
96237,
158155,
59520,
151403,
103420
]
},
{
"word": "alot",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
272,
8689,
7369,
8050,
852,
6204,
1461,
2139,
3206,
7096,
6823,
6703,
17483,
69560,
65866,
94152,
155121,
71240
]
},
{
"word": "yummy",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
29719,
22667,
26051,
12604,
11440,
67353,
48994,
53841,
48480,
67381,
89985,
158347,
109588,
211135,
185245,
179912,
144278,
88231
]
},
{
"word": "kinda",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
8163,
14351,
14542,
4587,
1800,
20020,
4337,
4612,
5789,
7680,
3099,
5747,
37892,
156961,
207106,
188539,
222333,
178987
]
},
{
"word": "whilst",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
14833,
24192,
152483,
211054,
40032,
69353,
3252,
4467,
52424,
47905,
109090,
38819,
117192,
88013,
63272,
44284,
75910,
74427
]
},
{
"word": "luckily",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
15138,
49610,
45375,
59651,
21617,
49294,
21386,
8157,
8438,
5945,
8820,
9538,
3602,
23038,
57957,
117689,
112685,
99937
]
}
],
"emergence": {
"position": 170,
"top1": "\n",
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
209665,
240578,
127229,
209878,
183635,
185769,
197110,
140705,
229435,
11746,
241261,
247864,
2572,
72917,
231212,
116147,
5525,
1
]
},
"scan": [],
"vanilla": {
"trajectories": [
{
"word": "anyways",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
184905,
168844,
132463,
154990,
138003,
132183,
106373,
196888,
75551,
75867,
164091,
99516,
149726,
180208,
166745,
153350,
184466,
175430
]
},
{
"word": "alot",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
123601,
140647,
115835,
123592,
75589,
84980,
56787,
74240,
216237,
242512,
228086,
160556,
110572,
68396,
41178,
53833,
113592,
71469
]
},
{
"word": "yummy",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
190882,
197351,
186307,
113534,
136583,
134891,
98053,
126002,
99474,
202876,
186873,
158539,
152819,
178540,
152632,
176997,
140765,
170763
]
},
{
"word": "kinda",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
96260,
130868,
83196,
73547,
37427,
35390,
76880,
183229,
55490,
65783,
74479,
23293,
138661,
232349,
225927,
211251,
239938,
229020
]
},
{
"word": "whilst",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
122067,
111899,
136962,
210395,
173232,
170455,
167857,
145686,
182169,
200093,
185170,
158854,
188715,
149595,
166127,
134582,
77201,
67688
]
},
{
"word": "luckily",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
156604,
155863,
147030,
123807,
103266,
90049,
59777,
52892,
59430,
37087,
56765,
25793,
4322,
7908,
150762,
182544,
181230,
131841
]
}
],
"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": 0.4536,
"32": 0.4536,
"36": 0.4536,
"40": 0.4536
}
]
}