INDEX

Explanations

phrases related to statements made by specific individuals

oai_token-act-pair · gpt-3.5-turbo

colons followed by statements or claims

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

 tremend

-0.90

inement

-0.73

 behavi

-0.73

ilater

-0.72

undai

-0.72

 pursu

-0.72

 behav

-0.70

 peac

-0.68

inarily

-0.67

 recip

-0.67

POSITIVE LOGITS

 Logged

0.93

TBD

0.87

 Provided

0.84

 Cosponsors

0.80

 Yeah

0.77

Who

0.75

ËĪ

0.75

 Rise

0.74

 Prelude

0.73

 Played

0.71

Activations Density 0.122%

phrases related to statements made by specific individuals

colons followed by statements or claims

No Comments

No Known Activations