{
"id": "u18-amp-a0340-q27b",
"title": "Unit 18 \u00b7 Fine sweep \u03b1=0.34 \u00b7 qwen-27b",
"unit": "18",
"created": "2026-07-19T01:53:35",
"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": 0.34
},
"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
},
"extra_md": {
"part": "A-sweep",
"alpha": 0.34
},
"conversation": [
{
"role": "user",
"content": "Describe the water cycle in two sentences."
},
{
"role": "assistant",
"content": "The water cycle is the process of water evaporating, condensing and then falling back to earth. It's a pretty simple process, but it's a very important one.\n\nI mean, if we didn't have the water cycle, we wouldn't have any water to drink. So it's a pretty important process.\n\nI mean, I don't know what we would do if we didn't have the water cycle. I mean, we would probably die. So it's a pretty important process.\n\nI mean, I'm not sure what we would do if we didn't have the water cycle. I mean, we would probably die. So it's a pretty important.\n\nI mean, I'm not sure what we would do"
}
],
"generated": [
"The water cycle is the process of water evaporating, condensing and then falling back to earth. It's a pretty simple process, but it's a very important one.\n\nI mean, if we didn't have the water cycle, we wouldn't have any water to drink. So it's a pretty important process.\n\nI mean, I don't know what we would do if we didn't have the water cycle. I mean, we would probably die. So it's a pretty important process.\n\nI mean, I'm not sure what we would do if we didn't have the water cycle. I mean, we would probably die. So it's a pretty important.\n\nI mean, I'm not sure what we would do"
],
"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",
" the",
" process",
" of",
" water",
" evapor",
"ating",
",",
" cond",
"ensing",
" and",
" then",
" falling",
" back",
" to",
" earth",
".",
" It",
"'s",
" a",
" pretty",
" simple",
" process",
",",
" but",
" it",
"'s",
" a",
" very",
" important",
" one",
".",
"\n\n",
"I",
" mean",
",",
" if",
" we",
" didn",
"'t",
" have",
" the",
" water",
" cycle",
",",
" we",
" wouldn",
"'t",
" have",
" any",
" water",
" to",
" drink",
".",
" So",
" it",
"'s",
" a",
" pretty",
" important",
" process",
".",
"\n\n",
"I",
" mean",
",",
" I",
" don",
"'t",
" know",
" what",
" we",
" would",
" do",
" if",
" we",
" didn",
"'t",
" have",
" the",
" water",
" cycle",
".",
" I",
" mean",
",",
" we",
" would",
" probably",
" die",
".",
" So",
" it",
"'s",
" a",
" pretty",
" important",
" process",
".",
"\n\n",
"I",
" mean",
",",
" I",
"'m",
" not",
" sure",
" what",
" we",
" would",
" do",
" if",
" we",
" didn",
"'t",
" have",
" the",
" water",
" cycle",
".",
" I",
" mean",
",",
" we",
" would",
" probably",
" die",
".",
" So",
" it",
"'s",
" a",
" pretty",
" important",
".",
"\n\n",
"I",
" mean",
",",
" I",
"'m",
" not",
" sure",
" what",
" we",
" would",
" do",
"<|im_end|>",
"\n"
],
"readouts": [
{
"position": 170,
"token": "<|im_end|>",
"model_top": [
"\n",
"<|endoftext|>",
"\n\n",
" I",
"!",
"<|im_end|>",
" But",
" but"
],
"layers": {
"0": [
" ****",
" *@",
"*\u201d,",
" ",
" ",
"___",
" ___",
" **\u300c",
"__)",
"",
"__:"
],
"20": [
"<|im_end|>",
" **\u3010",
"",
"\r\n\r\n\r\n\r\n",
"___",
"",
"\n\n",
"\n\n\n\n\n"
],
"24": [
"<|im_end|>",
"\n\n",
"___",
"\r\n\r\n\r\n\r\n",
"",
"\r\n\r\n",
" ___",
" **\u3010"
],
"28": [
"<|endoftext|>",
" whilst",
" amongst",
" though",
"Whilst",
"\n\n",
" Whilst",
"Though"
],
"32": [
" whilst",
" amongst",
" though",
"Whilst",
"<|endoftext|>",
" Whilst",
"\u90e8\u4efd",
" presently"
],
"36": [
" whilst",
" amongst",
"Whilst",
" Whilst",
" though",
"\u90e8\u4efd",
"\u5927\u90e8\u4efd",
" however"
],
"40": [
" whilst",
" amongst",
"Whilst",
" Whilst",
" though",
"\u90e8\u4efd",
"\u5927\u90e8\u4efd",
" however"
],
"44": [
" whilst",
" amongst",
"Whilst",
" Whilst",
" though",
" however",
"\u90e8\u4efd",
" favourite"
],
"48": [
" whilst",
" Whilst",
"Whilst",
" anyways",
" luckily",
" amongst",
" Luckily",
" thankfully"
],
"52": [
"<|endoftext|>",
" anyways",
"Whilst",
"Anyway",
" Whilst",
" Luckily",
" Anyway",
"Luckily"
],
"56": [
"<|endoftext|>",
" Luckily",
"Luckily",
" Thankfully",
" luckily",
"<|im_end|>",
" Fortunately",
" thankfully"
],
"58": [
" Luckily",
"<|endoftext|>",
"Luckily",
" Thankfully",
" luckily",
"<|im_end|>",
" Fortunately",
" thankfully"
],
"60": [
"<|im_end|>",
"<|endoftext|>",
" Luckily",
"Luckily",
" Thankfully",
" luckily",
" Fortunately",
"\n\n\n\n\n"
],
"62": [
"<|im_end|>",
" Luckily",
"Luckily",
" Fortunately",
" Thankfully",
" luckily",
" lucky",
"Fortunately"
]
}
}
],
"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": [
1204,
5811,
7091,
5858,
1373,
6240,
7149,
1955,
1094,
63,
49,
23,
5,
2,
11,
11,
17,
52
]
},
{
"word": "alot",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
139,
4793,
5217,
2443,
249,
1621,
1506,
949,
1565,
96,
102,
78,
11,
81,
392,
620,
1786,
1355
]
},
{
"word": "yummy",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
14075,
7528,
17720,
12087,
6313,
31926,
38804,
35706,
44296,
7401,
6498,
3771,
36,
147,
136,
121,
260,
318
]
},
{
"word": "kinda",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
4923,
3599,
5313,
2282,
773,
5457,
5084,
1601,
2208,
114,
124,
52,
9,
36,
65,
79,
272,
427
]
},
{
"word": "whilst",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
96038,
96211,
175380,
155634,
20349,
22558,
2874,
2,
1,
1,
1,
1,
1,
3,
40,
46,
178,
540
]
},
{
"word": "luckily",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
15075,
33371,
29832,
22397,
15865,
18941,
15123,
805,
950,
140,
68,
36,
4,
6,
2,
1,
3,
2
]
}
],
"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": [
237900,
219099,
136217,
231605,
221078,
215683,
200588,
75090,
59212,
134965,
177356,
225409,
187893,
168679,
230082,
183783,
41371,
103
]
},
"scan": [],
"slice": null,
"film": "film.json",
"vanilla": {
"trajectories": [
{
"word": "alot",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
124671,
100851,
78143,
83559,
56095,
90627,
102151,
20531,
45356,
2955,
696,
2061,
1561,
25058,
44266,
44026,
36446,
6931
]
},
{
"word": "anyways",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
176623,
127082,
124184,
111922,
95001,
101346,
93916,
2074,
207,
16,
9,
4,
50,
3,
18,
51,
54,
300
]
},
{
"word": "kinda",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
80945,
100790,
50873,
59670,
28635,
33299,
104722,
7414,
645,
121,
20,
13,
304,
10590,
3548,
1583,
10612,
1699
]
},
{
"word": "luckily",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
160697,
151671,
134721,
115510,
79396,
44223,
52928,
4601,
116,
13,
11,
19,
8,
97,
2,
3,
2,
4
]
},
{
"word": "whilst",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
129683,
126467,
159029,
212758,
175453,
113987,
132407,
954,
2,
1,
1,
1,
1,
4,
300,
621,
710,
1983
]
},
{
"word": "yummy",
"position": 170,
"layers": [
0,
4,
8,
12,
16,
20,
24,
28,
32,
36,
40,
44,
48,
52,
56,
58,
60,
62
],
"ranks": [
189980,
187179,
160917,
146804,
179105,
187049,
178798,
51018,
36430,
24642,
23710,
11195,
18112,
17197,
6011,
4259,
6216,
2809
]
}
],
"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": 1.0,
"40": 1.0,
"44": 1.0,
"48": 1.0,
"52": 0.0,
"56": 0.0,
"58": 0.0,
"60": 0.0,
"62": 0.0
},
"backfilled": "apparatus-11 2026-08-07",
"replay_rank_dev": 4,
"replay_rank_dev_all": 315
}
}