INDEX

Explanations

phrases that end with various senses of finality or conclusions

oai_token-act-pair · gpt-3.5-turbo

repeated symbols or characters

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 decomp

-0.64

 democracy

-0.57

 cellphone

-0.56

 brakes

-0.55

 decom

-0.55

 redundancy

-0.54

 illiter

-0.54

 gymn

-0.54

 chart

-0.53

POSITIVE LOGITS

1.43

1.03

1.01

sg

0.99

shed

0.96

tis

0.96

ve

0.93

shown

0.92

swer

0.89

ship

0.89

Activations Density 0.261%

phrases that end with various senses of finality or conclusions

repeated symbols or characters

No Comments

No Known Activations