INDEX

Explanations

numbers or numerical patterns related to information or statistics

oai_token-act-pair · gpt-3.5-turbo

negative phrases or references

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

ãĤ¯

-0.85

AVG

-0.82

ĸļ

-0.79

ãĤ¡

-0.74

 DRAG

-0.74

 Aether

-0.74

HCR

-0.73

ãĥ£

-0.69

 Cassidy

-0.68

achu

-0.66

POSITIVE LOGITS

eight

1.13

sized

1.11

year

1.07

percent

1.07

six

1.06

legged

1.06

degree

1.04

million

1.02

nine

1.01

second

1.01

Activations Density 0.065%

numbers or numerical patterns related to information or statistics

negative phrases or references

No Comments

No Known Activations