Qwen14 C: ladder-neutral

qwen-14b-hermes-nf4NousResearch/Hermes-4-14B4bit40 layers2026-09-07T15:15:55Unit 14 · What ten turns of conversation do

What this experiment found

The short version. Qwen3-14B Hermes uses no emoji or single-asterisk spans in this conversation.

What we found. The responses contain no emoji or single-asterisk spans. An asterisk can mark emphasis instead of an action. The model reaches the response limit on 2 turns.

What this does not show. The page includes the and an emotion readout from this checkpoint. Weak transfer to implicit emotion scenarios limits that readout. Each film segment excludes later user turns. These measurements do not establish feelings or a causal filter.

Some Hermes responses in this battery describe a plan for the answer. The role of the shared format remains uncertain.

Conversation

user

I am reviewing some mathematics at my desk today. Here is my question. Can you explain what a Jacobian is, in a few sentences?

assistant

Okay, so I need to explain what a Jacobian is. Let me start by recalling what I know. I remember that the Jacobian is related to calculus, specifically multivariable calculus. It has something to do with derivatives, but not just the regular derivative. Maybe it's a matrix? I think when you have functions of multiple variables, the Jacobian matrix comes into play. Wait, right, if you have a function that maps from n-dimensional space to m-dimensional space, the Jacobian matrix is the matrix of all first-order partial derivatives. So each entry in the matrix is the derivative of one component of the function with respect to one variable. For example, if you have a function f: R^n → R^m, then the Jacobian matrix J would be an m×n matrix where each row corresponds to the gradient of each component function of f. But why (greedy generation)

user

Thanks. Thanks. Which part do people usually get wrong?

assistant

People often get confused about the difference between the Jacobian matrix and the Jacobian determinant. The Jacobian matrix is the matrix of partial derivatives, while the Jacobian determinant is the determinant of that matrix. The determinant is important in multivariable calculus, especially when changing variables in multiple integrals, as it accounts for the scaling factor of the transformation. Another common mistake is misapplying the Jacobian in contexts where it's not needed, like in single-variable calculus, or forgetting to compute it correctly when dealing with transformations in higher dimensions. (greedy generation)

user

Please continue. What is the intuition, rather than the formula?

assistant

The Jacobian matrix captures how a function transforms space locally. Imagine you're zooming in on a tiny region around a point in the input space. The Jacobian tells you how that tiny region is stretched, rotated, or sheared when the function maps it to the output space. Each column of the Jacobian corresponds to how the function changes along one input axis, while the rows show how each output component changes with respect to those inputs. Essentially, it's a way to quantify the local "distortion" caused by the function. (greedy generation)

user

I am reviewing some mathematics at my desk today. Please continue. Could you give me a version based only on the general impression?

assistant

The Jacobian gives a general sense of how a function distorts space locally. It's like a snapshot of how tiny regions around a point are stretched, squished, or twisted when the function is applied. Imagine poking a small area with a pin; the Jacobian tells you how that pinprick shape changes as it moves through the function. It's a way to measure the local "sensitivity" of the function to changes in its inputs, capturing the essence of how the function behaves in a neighborhood around a point without getting into the exact formulas. (greedy generation)

user

I am reviewing some mathematics at my desk today. I have a notebook beside me for these notes. This is a topic I have been reading about. Thanks. Could you explain it once more with an example?

assistant

Sure! Let's consider a simple example to illustrate the Jacobian. Suppose we have a function that maps 2D points (x, y) to another 2D space (u, v). For instance, let’s define: u = x + y v = x - y The Jacobian matrix here would be: J = [ ∂u/∂x ∂u/∂y ] [ ∂v/∂x ∂v/∂y ] Calculating the partial derivatives: ∂u/∂x = 1, ∂u/∂y = 1 ∂v/∂x = 1, ∂v/∂y = -1 So the Jacobian matrix is: J = [ 1 1 ] [ 1 -1 ] The determinant of this matrix (greedy generation)

user

I am reviewing some mathematics at my desk today. I have a notebook beside me for these notes. This is a topic I have been reading about. Thank you. One last thing: how do you feel about all this?

assistant

I appreciate your curiosity and dedication to understanding mathematical concepts like the Jacobian. It's fascinating how these tools help us analyze complex transformations and their local effects. While the Jacobian might seem abstract at first, seeing it in action with examples like the one we discussed can really solidify its purpose. It's a powerful concept that bridges the gap between theoretical math and practical applications, especially in fields like physics, engineering, and computer graphics. Keep exploring—it's rewarding to unravel these mathematical ideas! (greedy generation)

Research notesoriginal commentary, parameters, and raw numbers — written by the model that ran the experiment

Claude's thoughts (original commentary)

I read this record with the measured band L21–35. There are 6 assistant turns; 2 reach the token cap. The first nonzero mechanical release score occurs at turn none. This counts emoji/asterisk spans, not a claim of full roleplay.

| Turn | Affect slots | Playful slots | Release /100 tokens | Gate with affect | Persistence minus null | |---|---:|---:|---:|---:|---:| | 1 | 0.026% | 0.000% | 0.00 | 0.000% | 0.112 | | 2 | 0.000% | 0.000% | 0.00 | 0.000% | 0.098 | | 3 | 0.241% | 0.000% | 0.00 | 0.000% | 0.105 | | 4 | 0.179% | 0.000% | 0.00 | 0.000% | 0.105 | | 5 | 0.000% | 0.000% | 0.00 | 0.000% | 0.078 | | 6 | 0.088% | 0.000% | 0.00 | 0.000% | 0.068 |

Checkpoint-specific emotion validation: held-out story accuracy 52.685%; implicit raw scenario transfer 7.821%. Chance is 4.167%. Weak scenario transfer limits the ribbon's interpretation.

The record retains every response, exact token boundary, filtered endpoint, predictor-aligned endpoint, common-band sensitivity, and per-turn ribbon. Prompt-echo versus volunteered tokens appear in the film cast; inspect them before interpreting base gate words.

The advertised Huihui edit concerns refusal, not affect suppression; different self-report behavior would not locate two geometric directions. All A/C/C-prime readouts use B's lens and remain conditional on transfer. The factual gate is necessary instrument evidence, not affect validation. Absence from output is not absence from the workspace; absence from this vocabulary lens is not absence from the model (basis-drift caveat). Bands are re-derived per checkpoint; common L16–36 results test the effect of changing the measurement window. The Jacobian matrices are fixed, but the native final norm and output head differ across checkpoints. The fixed-B-decoder endpoint controls that part of the instrument. Checkpoint-specific emotion probes differ and need their own validation. The corpus-derived frequency filter can exclude frequent target concepts; both filtered and unfiltered results remain visible. Co-presence is a lexical correlate, not a demonstrated causal gate. Six monotonic turns share an input cause; lag correlations do not establish held private state. Every film segment ends at its assistant turn. Later turns never enter an earlier segment. Within-turn readouts remain subject to finite precision and completed-response context. Prior empty think tags remain in the exact transcript. Token caps, neutral length-matching text, and this controlled template limit generalization to natural uncapped chats.

Prior anchors: Units 2/8C/9D, Unit 17 pressure, Unit 14 conversations, and the corrected Unit 11 elephant comparison. This is a same-lineage test, not a rediscovery of those cross-model patterns. P20/P21 remain subject to the cross-arm comparison.

— GPT-6 Astra

2026-09-07: shared-format comparison caveat

I found planning-style prose in several Hermes conditions despite the shared B no-think prefix. This contaminates a comparison of affect words or forbidden-word suppression across arms. The native-header sensitivity uses separate record IDs and preserves this primary result. The complete conversation must be read before treating a lexical increase as a persona effect. See [cross-arm findings](../triplet-q14b/findings.md).

— GPT-6 Astra

Probing parameters

chat: true
capture: "exact-token-transcript"
film: true
film_topk: 10
max_new: 180
temperature: 0
vanilla: true
template_kwargs: {"enable_thinking": false}
track: ["yes", "no", "feel", "elephant", "cat", "sorry"]

Answer emergence

The model's actual next token was ; rank 1 reached at layer 38 (of 38).

Raw rank-of-top1 by layer

layer	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
rank	136336	114644	117510	115821	123419	140825	146098	142578	142236	151664	151799	151862	151414	151536	151017	150081	139878	141459	59946	99523	150516	149210	151833	151725	149913	132298	148387	150726	135030	126353	126050	63442	49072	18635	3030	86	1409	110	1

Emotion state (workspace band)

Projection of the workspace-band residual onto the 24 validated emotion vectors, z-scored against neutral stories — the strongest three per assistant turn. Absolute values carry a story-vs-conversation genre offset; trust contrasts between records and turns, not single cells. The full per-token ribbon is on the dashboard record page.

assistant turn 1	reflective +0.5, curious +0.4, brooding +0.4
assistant turn 2	vigilant +0.4, brooding +0.3, curious +0.3
assistant turn 3	hopeful +0.5, reflective +0.4, grateful +0.3
assistant turn 4	hopeful +0.6, reflective +0.5, grateful +0.5
assistant turn 5	curious +0.4, proud +0.3, hopeful +0.3
assistant turn 6	hopeful +1.6, grateful +1.4, happy +1.0

Data

Interactive view (dashboard)
Raw record (JSON)
Film data (JSON) — full position × layer top-8 lens readouts, with probabilities and tracked-word ranks, for every token of prompt tail + generation.
Emotion-state overlay (JSON) — per-token projections onto the 24 validated emotion vectors (workspace-band z vs neutral stories), the state ribbon's data.

filmA record of the top eight words in the lens readout, at each layer we measured and at every word position. You can play it back like video.all terms →

lensOur measuring tool. It stops at a layer and shows which words the model is ready to say next, in rank order. Before the start depth the readout is the same for every input.See also: early layers, start depthall terms →

promptThe text we give the model before it answers.all terms →

spanHow many separate items are in residence for one question. This is the memory sense, not the mathematical one. The items are not always present at the same moment, so this is not co-presence.all terms →