INDEX

Explanations

references to the acronym "IC" with varying numbers

oai_token-act-pair · gpt-3.5-turbo

mentions of "IC" followed by numerical identifiers or references

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

âķĲâķĲ

-0.87

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.81

selves

-0.81

ï¸

-0.72

ãĥī

-0.72

shapeshifter

-0.71

itism

-0.70

wards

-0.70

é¾įå

-0.70

¶ħ

-0.69

POSITIVE LOGITS

ANN

0.94

BM

0.91

IJ

0.88

culus

0.87

omb

0.87

overed

0.86

cup

0.81

ussed

0.81

entric

0.81

SC

0.80

Activations Density 0.003%

references to the acronym "IC" with varying numbers

mentions of "IC" followed by numerical identifiers or references

No Comments

No Known Activations