INDEX

Explanations

phrases related to legal, bureaucratic, or administrative processes

oai_token-act-pair · gpt-3.5-turbo

the word "the" in connection with data references

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

Allows

-0.71

ÃĽ

-0.66

LY

-0.65

otto

-0.65

arella

-0.63

 Akin

-0.61

 Cumm

-0.61

ãĥ¯

-0.61

 Mori

-0.60

uca

-0.60

POSITIVE LOGITS

 foregoing

0.90

 aforementioned

0.89

 above

0.73

 sexes

0.72

ses

0.70

gebra

0.70

above

0.69

 hairs

0.69

 digits

0.68

 principals

0.68

Activations Density 0.135%

phrases related to legal, bureaucratic, or administrative processes

the word "the" in connection with data references

No Comments

No Known Activations