INDEX

Explanations

sentences indicating accusations or criticism

oai_token-act-pair · gpt-3.5-turbo

instances of accusations or blame

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

abouts

-0.76

ãĥ¼ãĤ¯

-0.75

ï¸

-0.74

PUT

-0.73

ãĤ¹ãĥĪ

-0.70

ching

-0.68

ÃįÃį

-0.65

ÏĦ

-0.65

hops

-0.65

ãĥ©ãĥ³

-0.64

POSITIVE LOGITS

 veto

0.65

ITV

0.63

 Kinn

0.62

 hosp

0.61

 motto

0.61

 Netherlands

0.58

anca

0.57

 Luxembourg

0.57

 intestine

0.57

 gulf

0.57

Activations Density 0.000%

sentences indicating accusations or criticism

instances of accusations or blame

No Comments

No Known Activations