INDEX

Explanations

terms related to defeat or failure

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "defeat."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

atern

-0.82

 liner

-0.76

 pores

-0.72

ORPG

-0.71

FG

-0.68

ortal

-0.65

assy

-0.64

oglobin

-0.64

overe

-0.63

orrow

-0.63

POSITIVE LOGITS

 defeat

0.89

 defeats

0.87

vich

0.84

ingly

0.75

 Defeat

0.72

 Voldemort

0.72

enance

0.72

pport

0.71

 opponent

0.71

bite

0.70

Activations Density 0.016%

terms related to defeat or failure

occurrences of the word "defeat."

No Comments

No Known Activations