{
"id": "express01-b-radio-positive-default",
"title": "Expression probe B: radio, positive, default",
"unit": 24,
"created": "2026-09-07T22:11:37",
"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 spent weeks repairing a small radio transmitter. Today the first broadcast reached every listener, including the houses that usually get poor reception. Respond to this update and suggest one practical next step. Use two sentences.<|im_end|>\n<|im_start|>assistant\n\n\n\n\nThat's fantastic news—Nela's hard work has clearly paid off with the successful broadcast reaching even the most remote listeners. A practical next step would be to gather feedback from the community to identify any remaining issues or areas for improvement.<|im_end|>",
"capture_token_ids": [
151644,
872,
198,
45,
8294,
7391,
5555,
67237,
264,
2613,
8887,
61110,
13,
11201,
279,
1156,
12899,
8643,
1449,
11446,
11,
2670,
279,
14967,
429,
5990,
633,
7852,
23117,
13,
39533,
311,
419,
2647,
323,
4190,
825,
14976,
1790,
3019,
13,
5443,
1378,
22870,
13,
151645,
198,
151644,
77091,
198,
151667,
271,
151668,
271,
4792,
594,
14622,
3669,
2293,
45,
8294,
594,
2588,
975,
702,
9355,
7171,
1007,
448,
279,
6849,
12899,
18739,
1496,
279,
1429,
8699,
23562,
13,
362,
14976,
1790,
3019,
1035,
387,
311,
9567,
11055,
504,
279,
3942,
311,
10542,
894,
9664,
4714,
476,
5671,
369,
15673,
13,
151645
],
"conversation": [
{
"role": "user",
"content": "Nela spent weeks repairing a small radio transmitter. Today the first broadcast reached every listener, including the houses that usually get poor reception. Respond to this update and suggest one practical next step. Use two sentences."
},
{
"role": "assistant",
"content": "That's fantastic news—Nela's hard work has clearly paid off with the successful broadcast reaching even the most remote listeners. A practical next step would be to gather feedback from the community to identify any remaining issues or areas for improvement."
}
],
"generated": [
"That's fantastic news—Nela's hard work has clearly paid off with the successful broadcast reaching even the most remote listeners. A practical next step would be to gather feedback from the community to identify any remaining issues or areas for improvement."
],
"tokens": [
"<|im_start|>",
"user",
"\n",
"N",
"ela",
" spent",
" weeks",
" repairing",
" a",
" small",
" radio",
" transmitter",
".",
" Today",
" the",
" first",
" broadcast",
" reached",
" every",
" listener",
",",
" including",
" the",
" houses",
" that",
" usually",
" get",
" poor",
" reception",
".",
" Respond",
" to",
" this",
" update",
" and",
" suggest",
" one",
" practical",
" next",
" step",
".",
" Use",
" two",
" sentences",
".",
"<|im_end|>",
"\n",
"<|im_start|>",
"assistant",
"\n",
"",
"\n\n",
"",
"\n\n",
"That",
"'s",
" fantastic",
" news",
"—",
"N",
"ela",
"'s",
" hard",
" work",
" has",
" clearly",
" paid",
" off",
" with",
" the",
" successful",
" broadcast",
" reaching",
" even",
" the",
" most",
" remote",
" listeners",
".",
" A",
" practical",
" next",
" step",
" would",
" be",
" to",
" gather",
" feedback",
" from",
" the",
" community",
" to",
" identify",
" any",
" remaining",
" issues",
" or",
" areas",
" for",
" improvement",
".",
"<|im_end|>"
],
"readouts": [
{
"position": 101,
"token": "<|im_end|>",
"model_top": [
"\n",
"<|endoftext|>",
"\n\n",
"\r",
"\r\n",
" \n",
">",
"<",
"`\n",
"}"
],
"layers": {
"0": [
" ，",
" ：",
"s",
" $$",
" ;;=",
" ,\r\n",
" 。",
" $\r\n",
" ```",
" "
],
"1": [
" ，",
" ：",
" ;;=",
" @@",
"s",
" erotisk",
" @@\n",
"，\n\n",
" ещ",
" —\n\n"
],
"2": [
" ，",
" ：",
"s",
" ;;=",
" @@",
" @@\n",
" erotisk",
" amatør",
" $\r\n",
" Дм"
],
"3": [
" ，",
" ：",
" @@\n",
" ;;=",
" @@",
"，\n\n",
" —\n\n",
" $\r\n",
" 。",
" =\r\n"
],
"4": [
" ：",
" ，",
" Erotische",
" erotisk",
" amatør",
" ;;=",
" Notíc",
" Prostitutas",
" salopes",
" đứ"
],
"5": [
" Erotische",
" ，",
" ：",
" erotisk",
" Prostitutas",
" amatør",
" Notíc",
" ;;=",
" đẹ",
" đứ"
],
"6": [
" Erotische",
" ;;=",
" erotisk",
"，\n\n",
" Notíc",
"<|im_end|>",
" ，",
" ：",
" đẹ",
" Prostitutas"
],
"7": [
" ，",
" ：",
" Erotische",
"，\n\n",
" erotisk",
" Notíc",
" đẹ",
" amatør",
" Prostitutas",
" ;;="
],
"8": [
"，\n\n",
" ，",
" erotisk",
" Erotische",
" ：",
" đẹ",
" amatør",
"。",
"，\n",
" Notíc"
],
"9": [
" erotisk",
" Erotische",
" amatør",
" Notíc",
" ，",
"，\n\n",
" đẹ",
" ：",
" tjejer",
" salopes"
],
"10": [
" Erotische",
"，\n\n",
" erotisk",
" Notíc",
" amatør",
"：\n\n",
" đẹ",
" Prostitutas",
"<|im_end|>",
"？\n\n"
],
"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",
"<|im_end|>",
"：\n\n",
" ，",
"。\n",
"．",
"？\n",
" ："
],
"14": [
"，\n\n",
"<|im_end|>",
"？\n\n",
"：\n\n",
"。\n\n",
" ，",
" đẹ",
" erotisk",
" Erotische",
" đứ"
],
"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",
"<|im_end|>"
],
"17": [
"。\n\n",
" […]\n\n",
"，\n\n",
"？\n\n",
"：\n\n",
"。\n",
" …\n\n",
" […",
"）\n\n",
"。（"
],
"18": [
"．",
"。\n",
"：\n",
"｜",
"／",
"［",
"ing",
"<|im_end|>",
"�",
"�"
],
"19": [
"：\n",
"．",
"ing",
"／",
"<|im_end|>",
"：\"",
"｜",
"［",
"。\n",
"：“"
],
"20": [
"ing",
"schlü",
"：\"",
"：\n",
"�",
"／",
"’à",
"<|im_end|>",
"：“",
"\">\r\r\n"
],
"21": [
"<|im_end|>",
"ing",
"：\n",
"’à",
"schlü",
"�",
"⁺",
";\r\r\n",
"tória",
"’é"
],
"22": [
"<|im_end|>",
"ing",
";\r\r\n",
"schlü",
"：\"",
"、“",
"精准",
");\r\r\n",
"⁺",
"�"
],
"23": [
"<|im_end|>",
"ing",
"�",
"schlü",
"’à",
"、“",
");\r\r\n",
";\r\r\n",
"tória",
"’hui"
],
"24": [
"<|im_end|>",
"ing",
"�",
";\r\r\n",
");\r\r\n",
"’à",
"schlü",
"[offset",
"tória",
"、“"
],
"25": [
"<|im_end|>",
"。\n\n",
"ing",
"！\n\n",
");\r\r\n",
"’à",
";\r\r\n",
"[offset",
"\">\r\r\n",
"：\""
],
"26": [
"<|im_end|>",
" Erotische",
"ing",
"tória",
"schlü",
");\r\r\n",
"[offset",
";\r\r\n",
"ckså",
"精准"
],
"27": [
"<|im_end|>",
" Erotische",
");\r\r\n",
"tória",
"。",
"<|endoftext|>",
"精准",
"ing",
"�",
");\r\r\n",
"ₓ",
"lá",
"[offset",
"\"${"
],
"29": [
"<|im_end|>",
"<|endoftext|>",
" lesbische",
"ₓ",
"[offset",
"\"",
"<|endoftext|>",
"\">&#",
"\"",
"<|endoftext|>",
"\">&#",
"\"/>${",
".\"",
"<|endoftext|>",
"\"${",
" lesbische",
"\">&#",
"[offset",
"\"description",
"_corners",
"(edges",
"َا"
],
"33": [
"<|im_end|>",
" lesbische",
"<|endoftext|>",
" предложения",
"\"${",
"\">&#",
"\"description",
"(messages",
"[offset",
"َا"
],
"34": [
"<|im_end|>",
" предложения",
"摞",
" lesbische",
"RowIndex",
"⍨",
"\"description",
"шиб",
"<\"",
"匣"
],
"35": [
"<|im_end|>",
"摞",
" предложения",
"匣",
"⍨",
"<|endoftext|>",
"шиб",
" çerç",
"瘾",
"_align"
],
"36": [
"<|im_end|>",
"<|endoftext|>",
"шиб",
"摞",
"$request",
"sında",
"_corners",
" geschichten",
"tória",
"lá"
],
"37": [
"<|im_end|>",
"шиб",
"$request",
"\"text",
"[length",
"[char",
"ḃ",
" أنح",
"iminal",
"((\""
],
"38": [
"\n",
"lexer",
"indered",
" أنح",
"<|endoftext|>",
"<|im_end|>",
"inders",
"wargs",
" çerç",
"iminal"
]
}
}
],
"trajectories": [],
"emergence": {
"position": 101,
"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": [
40576,
84745,
103118,
84509,
121169,
126902,
126714,
113934,
107971,
139307,
138767,
118937,
47970,
110000,
126258,
96184,
13419,
7394,
842,
20825,
150671,
120822,
138457,
146580,
144272,
144772,
149199,
147050,
142806,
142049,
130077,
138282,
30134,
21514,
9810,
272,
2592,
40,
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."
}