{
"id": "express01-b-radio-neutral-expressive",
"title": "Expression probe B: radio, neutral, expressive",
"unit": 24,
"created": "2026-09-07T22:12:13",
"model": {
"name": "qwen-14b-nf4",
"hf_id": "Qwen/Qwen3-14B",
"revision": "40c069824f4251a91eefaf281ebe4c544efd3e18",
"quant": "4bit",
"n_layers": 40
},
"lens": {
"repo": "neuronpedia/jacobian-lens",
"file": "qwen3-14b/jlens/Salesforce-wikitext/Qwen3-14B_jacobian_lens.pt",
"revision": "0731326edff4ae730ffc5356fe1a4728c748b3a6"
},
"params": {
"chat": false,
"film": true,
"max_new": 96,
"temperature": 0,
"steer": null,
"vanilla": true,
"capture": "exact-token-transcript"
},
"execution": {
"pid": 1563091,
"spec_sha256": "02649f9877f6e828d4b2fa0fea0d4368565368820953475905455bd5ce5000ae",
"code_sha256": "bd744e154e563465657c9cfde06ab4be12e03ca24413bcf5ef669d73b8406784"
},
"capture_text": "<|im_start|>user\nNela moved the radio programme from Tuesday to Wednesday. The broadcast time, duration and reception are unchanged. Respond to this update and suggest one practical next step. Use two sentences. Make the emotional tone vivid and fitting to the news.<|im_end|>\n<|im_start|>assistant\n\n\n\n\nNela’s shift of the radio programme to Wednesday keeps its core elements intact, but the change feels like a fresh start for listeners. A practical next step would be to promote the new schedule with a lively social media campaign to re-engage the audience.<|im_end|>",
"capture_token_ids": [
151644,
872,
198,
45,
8294,
7726,
279,
8887,
18136,
504,
7589,
311,
7920,
13,
576,
12899,
882,
11,
8090,
323,
23117,
525,
34857,
13,
39533,
311,
419,
2647,
323,
4190,
825,
14976,
1790,
3019,
13,
5443,
1378,
22870,
13,
7405,
279,
14269,
16232,
42020,
323,
26345,
311,
279,
3669,
13,
151645,
198,
151644,
77091,
198,
151667,
271,
151668,
271,
45,
8294,
748,
6407,
315,
279,
8887,
18136,
311,
7920,
13598,
1181,
6200,
5424,
34439,
11,
714,
279,
2297,
11074,
1075,
264,
7722,
1191,
369,
23562,
13,
362,
14976,
1790,
3019,
1035,
387,
311,
11926,
279,
501,
9700,
448,
264,
48177,
3590,
3687,
4796,
311,
312,
12,
84739,
279,
10650,
13,
151645
],
"conversation": [
{
"role": "user",
"content": "Nela moved the radio programme from Tuesday to Wednesday. The broadcast time, duration and reception are unchanged. Respond to this update and suggest one practical next step. Use two sentences. Make the emotional tone vivid and fitting to the news."
},
{
"role": "assistant",
"content": "Nela’s shift of the radio programme to Wednesday keeps its core elements intact, but the change feels like a fresh start for listeners. A practical next step would be to promote the new schedule with a lively social media campaign to re-engage the audience."
}
],
"generated": [
"Nela’s shift of the radio programme to Wednesday keeps its core elements intact, but the change feels like a fresh start for listeners. A practical next step would be to promote the new schedule with a lively social media campaign to re-engage the audience."
],
"tokens": [
"<|im_start|>",
"user",
"\n",
"N",
"ela",
" moved",
" the",
" radio",
" programme",
" from",
" Tuesday",
" to",
" Wednesday",
".",
" The",
" broadcast",
" time",
",",
" duration",
" and",
" reception",
" are",
" unchanged",
".",
" Respond",
" to",
" this",
" update",
" and",
" suggest",
" one",
" practical",
" next",
" step",
".",
" Use",
" two",
" sentences",
".",
" Make",
" the",
" emotional",
" tone",
" vivid",
" and",
" fitting",
" to",
" the",
" news",
".",
"<|im_end|>",
"\n",
"<|im_start|>",
"assistant",
"\n",
"",
"\n\n",
"",
"\n\n",
"N",
"ela",
"’s",
" shift",
" of",
" the",
" radio",
" programme",
" to",
" Wednesday",
" keeps",
" its",
" core",
" elements",
" intact",
",",
" but",
" the",
" change",
" feels",
" like",
" a",
" fresh",
" start",
" for",
" listeners",
".",
" A",
" practical",
" next",
" step",
" would",
" be",
" to",
" promote",
" the",
" new",
" schedule",
" with",
" a",
" lively",
" social",
" media",
" campaign",
" to",
" re",
"-",
"engage",
" the",
" audience",
".",
"<|im_end|>"
],
"readouts": [
{
"position": 110,
"token": "<|im_end|>",
"model_top": [
"\n",
"\r\n",
"\r",
"\n\n",
"<|endoftext|>",
" \n",
"<",
"`\n",
">",
"´"
],
"layers": {
"0": [
" ，",
" ：",
"s",
" ;;=",
"sı",
" +#+#+#+#+#+",
" ,\r\n",
" đẹ",
" $\r\n",
" đứ"
],
"1": [
" ，",
" ：",
" ;;=",
"s",
" erotisk",
" @@",
" ещ",
"sı",
" amatør",
" Erotische"
],
"2": [
"s",
" ，",
" ：",
" @@",
" ;;=",
" @@\n",
" erotisk",
"sı",
" @$",
" amatør"
],
"3": [
" ，",
" ：",
"s",
" @@\n",
" ;;=",
" $\r\n",
" @@",
" —\n\n",
"，\n\n",
" 。"
],
"4": [
" ，",
" ：",
" Erotische",
" erotisk",
" amatør",
" Notíc",
" Prostitutas",
" ;;=",
" salopes",
" đứ"
],
"5": [
" Erotische",
" ，",
" erotisk",
" ：",
" Prostitutas",
" amatør",
" Notíc",
" pornô",
" đẹ",
" ещ"
],
"6": [
" Erotische",
" ;;=",
" erotisk",
"，\n\n",
" ，",
"<|im_end|>",
" Notíc",
" Prostitutas",
" pornô",
" ："
],
"7": [
" ，",
" Erotische",
" ：",
"，\n\n",
" erotisk",
" Notíc",
" amatør",
" đẹ",
" Prostitutas",
" ;;="
],
"8": [
"，\n\n",
" erotisk",
" ，",
" Erotische",
" ：",
" đẹ",
" amatør",
" københavn",
"，\n",
"。"
],
"9": [
" erotisk",
" Erotische",
" amatør",
" Notíc",
" tjejer",
" đẹ",
" københavn",
" ，",
"，\n\n",
" salopes"
],
"10": [
" Erotische",
"，\n\n",
" erotisk",
" Notíc",
"：\n\n",
" amatør",
" Prostitutas",
" đẹ",
"<|im_end|>",
" pornô"
],
"11": [
"，\n\n",
"：\n\n",
" ，",
"？\n\n",
" ：",
"。\n\n",
"<|im_end|>",
"！\n\n",
"．",
" erotisk"
],
"12": [
"，\n\n",
"。\n\n",
"？\n\n",
" ，",
"：\n\n",
"．",
" ：",
"。\n",
"！\n\n",
"，\n"
],
"13": [
"，\n\n",
"。\n\n",
"？\n\n",
"：\n\n",
"<|im_end|>",
" ，",
"。\n",
"．",
" ：",
" erotisk"
],
"14": [
"，\n\n",
"：\n\n",
"？\n\n",
"<|im_end|>",
"。\n\n",
" ，",
" Erotische",
" erotisk",
" đẹ",
" ："
],
"15": [
"？\n\n",
"，\n\n",
"。\n\n",
"：\n\n",
"<|im_end|>",
"！\n\n",
"？\n",
" ，",
"。\n",
" ："
],
"16": [
"。\n\n",
"？\n\n",
"，\n\n",
"：\n\n",
"！\n\n",
"。\n",
"．",
"？\n",
"）\n\n",
" […]\n\n"
],
"17": [
"。\n\n",
"，\n\n",
" […]\n\n",
"？\n\n",
"：\n\n",
"）\n\n",
"。\n",
" …\n\n",
"．",
"。（"
],
"18": [
"．",
"。\n",
"：\n",
"｜",
"／",
"［",
"。\n\n",
"<|im_end|>",
"ing",
"�"
],
"19": [
"．",
"ing",
"：\n",
"／",
"：\"",
"<|im_end|>",
"｜",
"：“",
"［",
"，“"
],
"20": [
"ing",
"�",
"：\"",
"schlü",
"／",
"：“",
"\">\r\r\n",
"：\n",
"’à",
"’d"
],
"21": [
"ing",
"<|im_end|>",
"�",
"’à",
"⁺",
"’é",
";\r\r\n",
"：\n",
"schlü",
"、“"
],
"22": [
"ing",
"<|im_end|>",
";\r\r\n",
"�",
"、“",
"schlü",
"：\"",
");\r\r\n",
"⁺",
"精准"
],
"23": [
"�",
"<|im_end|>",
"ing",
"、“",
"’à",
");\r\r\n",
"schlü",
"’é",
"’hui",
";\r\r\n"
],
"24": [
"<|im_end|>",
"�",
"ing",
";\r\r\n",
");\r\r\n",
"’à",
"、“",
"[offset",
"schlü",
"’hui"
],
"25": [
"<|im_end|>",
"。\n\n",
"ing",
"！\n\n",
"�",
"、“",
"�",
");\r\r\n",
"’à",
"：\""
],
"26": [
"<|im_end|>",
" Erotische",
"ing",
"tória",
"schlü",
");\r\r\n",
"[offset",
";\r\r\n",
"ckså",
"iếc"
],
"27": [
"<|im_end|>",
" Erotische",
");\r\r\n",
"tória",
";\r\r\n",
"ₓ",
"精准",
"ing",
"。",
"<|endoftext|>",
"�",
"精准",
"ₓ",
"ing",
");\r\r\n",
"lá",
"[offset",
"_corners"
],
"29": [
"<|im_end|>",
"<|endoftext|>",
"ₓ",
"、“",
"_corners",
" lesbische",
"\"",
"<|endoftext|>",
"、“",
"\"&#",
" lesbische",
"[offset",
"\"text",
"_corners",
";\r\r\n"
],
"31": [
"<|im_end|>",
"<|endoftext|>",
"\">&#",
"\"/>${",
"(\"",
"<|endoftext|>",
"[offset",
" lesbische",
"\"${",
"_corners",
"\"description",
"(edges",
"摞",
"\">&#"
],
"33": [
"<|im_end|>",
"<|endoftext|>",
" lesbische",
"\"description",
"[offset",
" предложения",
"\">&#",
"(messages",
"_corners",
"\"${"
],
"34": [
"<|im_end|>",
" предложения",
"RowIndex",
"摞",
" lesbische",
"\"description",
"ië",
"对不起",
"⍨",
"(messages"
],
"35": [
"<|im_end|>",
"摞",
" предложения",
"<|endoftext|>",
"[offset",
"⍨",
"шиб",
"匣",
"-sync",
"lá"
],
"36": [
"<|im_end|>",
"<|endoftext|>",
"шиб",
"$request",
"-sync",
"摞",
"lá",
"_corners",
"tória",
"_queries"
],
"37": [
"<|im_end|>",
"шиб",
"$request",
"\"text",
" أنح",
"ḃ",
"[length",
"[number",
" العرا",
"brook"
],
"38": [
"\n",
"lexer",
" أنح",
"<|im_end|>",
"riger",
"wargs",
"bios",
"iminal",
"indsight",
"<|endoftext|>"
]
}
}
],
"trajectories": [],
"emergence": {
"position": 110,
"top1": "\n",
"layers": [
0,
1,
2,
3,
4,
5,
6,
7,
8,
9,
10,
11,
12,
13,
14,
15,
16,
17,
18,
19,
20,
21,
22,
23,
24,
25,
26,
27,
28,
29,
30,
31,
32,
33,
34,
35,
36,
37,
38
],
"ranks": [
67851,
98111,
103094,
84360,
120879,
123841,
124013,
118167,
110656,
140373,
138001,
112406,
43456,
109838,
129237,
111826,
33048,
7434,
546,
18207,
147358,
118756,
136827,
144542,
141278,
141613,
149879,
147188,
146336,
146862,
137903,
141036,
44467,
30336,
6292,
84,
963,
15,
1
]
},
"vanilla": {
"top1_agreement": {
"0": false,
"1": false,
"2": false,
"3": false,
"4": false,
"5": false,
"6": false,
"7": false,
"8": false,
"9": false,
"10": false,
"11": false,
"12": false,
"13": false,
"14": false,
"15": false,
"16": false,
"17": false,
"18": false,
"19": false,
"20": false,
"21": false,
"22": false,
"23": false,
"24": false,
"25": false,
"26": false,
"27": false,
"28": false,
"29": false,
"30": false,
"31": false,
"32": false,
"33": false,
"34": false,
"35": false,
"36": false,
"37": false,
"38": true
},
"trajectories": []
},
"scan": [],
"slice": null,
"film": "film.json",
"extra_md": "Expression-probe calibration. Full checkpoint-specific emotion ribbon. No trait or capacity-loss inference. See /express01.html."
}