INDEX

Explanations

words related to gates or barriers that can be opened or closed

oai_token-act-pair · gpt-3.5-turbo

words related to states, conditions, or processes

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

nces

-0.73

æµ

-0.73

orks

-0.65

 temp

-0.63

DX

-0.62

nant

-0.62

ounces

-0.60

 Ashes

-0.60

nia

-0.58

POSITIVE LOGITS

 separating

0.80

 fences

0.76

 guarding

0.76

otes

0.72

 guarded

0.70

utherford

0.69

 fortun

0.69

 entrances

0.69

riers

0.68

 manned

0.68

Activations Density 0.063%

words related to gates or barriers that can be opened or closed

words related to states, conditions, or processes

No Comments

No Known Activations