INDEX

Explanations

phrases indicating correctness or approval

oai_token-act-pair · gpt-3.5-turbo

expressions related to the concept of "rightness."

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

ĸļ

-0.81

mat

-0.71

ains

-0.69

ipation

-0.68

ulz

-0.65

cit

-0.64

ripp

-0.63

igmat

-0.61

graph

-0.61

 Railroad

-0.60

POSITIVE LOGITS

eous

1.30

 wing

0.82

 winger

0.80

wing

0.78

shore

0.77

 aligned

0.76

ward

0.65

 fielder

0.65

move

0.65

å¾

0.64

Activations Density 0.041%

phrases indicating correctness or approval

expressions related to the concept of "rightness."

No Comments

No Known Activations