INDEX

Explanations

words related to expressing gratitude or understanding

oai_token-act-pair · gpt-3.5-turbo

terms related to apologies and appreciation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ersion

-0.93

agus

-0.81

enegger

-0.78

UNE

-0.75

UGC

-0.75

Increase

-0.74

Ranked

-0.73

define

-0.71

 Minotaur

-0.71

RTX

-0.71

POSITIVE LOGITS

 apolog

3.14

 appreci

1.81

apolog

1.59

 stal

1.46

 desp

1.38

 irre

1.25

coy

1.17

 strateg

1.10

 congrat

1.06

 opportun

1.04

Activations Density 0.048%

words related to expressing gratitude or understanding

terms related to apologies and appreciation

No Comments

No Known Activations