INDEX

Explanations

references to locations and time periods

oai_token-act-pair · gpt-3.5-turbo

instances of emphasis or concern in statements

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

çīĪ

-0.74

cffffcc

-0.69

ãĤ´ãĥ³

-0.69

emi

-0.64

chieve

-0.64

undrum

-0.63

 reportedly

-0.63

APTER

-0.62

 commented

-0.61

 confir

-0.61

POSITIVE LOGITS

 "âĢ¦

0.86

 nobody

0.78

there

0.78

"'

0.77

they

0.77

"[

0.74

 they

0.71

 there

0.71

 terrorists

0.70

 attackers

0.70

Activations Density 0.384%

references to locations and time periods

instances of emphasis or concern in statements

No Comments

No Known Activations