INDEX

Explanations

statements where individuals are giving accounts of events or incidents involving themselves or others

oai_token-act-pair · gpt-3.5-turbo

pronouns indicating individuals involved in allegations of misconduct or personal experiences

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

focus

-0.71

 Nationwide

-0.70

catentry

-0.64

 Mandatory

-0.63

 implication

-0.62

 bureaucr

-0.62

 Leadership

-0.62

 Globe

-0.62

 newcom

-0.61

Official

-0.60

POSITIVE LOGITS

 woke

1.17

 slept

1.17

 awoke

1.15

 regretted

1.14

 cheated

1.13

 faked

1.12

 dreamed

1.10

 raped

1.10

 smelled

1.09

 drank

1.07

Activations Density 0.169%

statements where individuals are giving accounts of events or incidents involving themselves or others

pronouns indicating individuals involved in allegations of misconduct or personal experiences

No Comments

No Known Activations