INDEX

Explanations

phrases or words related to criminal investigations or charges

oai_token-act-pair · gpt-3.5-turbo

phrases indicating a relationship to legal investigations or charges

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

stall

-0.84

 preached

-0.74

ipel

-0.71

inse

-0.70

mare

-0.69

uck

-0.69

avorite

-0.67

 Fernand

-0.66

fl

-0.65

otted

-0.65

POSITIVE LOGITS

 Cases

0.74

 ties

0.74

 instances

0.74

 thereto

0.74

 connection

0.73

rals

0.71

 cases

0.71

 dealings

0.71

 robberies

0.70

ance

0.69

Activations Density 0.021%

phrases or words related to criminal investigations or charges

phrases indicating a relationship to legal investigations or charges

No Comments

No Known Activations