INDEX

Explanations

phrases related to taking action or making a stand

oai_token-act-pair · gpt-3.5-turbo

repeated conjunctions and calls to action

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

wic

-0.77

NULL

-0.71

notations

-0.70

nc

-0.68

Reviewer

-0.67

unique

-0.67

Asset

-0.67

far

-0.67

 Weekly

-0.66

iment

-0.66

POSITIVE LOGITS

 realize

0.88

 reap

0.87

 hopefully

0.87

 declare

0.86

 make

0.84

 enjoy

0.84

 forth

0.84

 bask

0.83

 apologize

0.83

let

0.81

Activations Density 0.168%

phrases related to taking action or making a stand

repeated conjunctions and calls to action

No Comments

No Known Activations