INDEX

Explanations

timestamps or temporal references

oai_token-act-pair · gpt-3.5-turbo

instances of punctuation or commas

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

SourceFile

-0.83

MpServer

-0.80

ļéĨĴ

-0.76

swick

-0.73

ãĤ´ãĥ³

-0.69

Done

-0.65

sequently

-0.63

iffe

-0.63

anz

-0.63

utive

-0.61

POSITIVE LOGITS

 however

0.99

we

0.89

 everybody

0.87

 expect

0.87

 everyone

0.86

 there

0.84

 nobody

0.80

 whenever

0.77

 unless

0.75

 everything

0.71

Activations Density 0.310%

timestamps or temporal references

instances of punctuation or commas

No Comments

No Known Activations