INDEX

Explanations

references to all-time highs and lows

oai_token-act-pair · gpt-3.5-turbo

references to "all-time highs" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Mind

-0.66

ournal

-0.65

rosis

-0.64

liga

-0.63

REDACTED

-0.63

rogen

-0.61

rium

-0.61

ropri

-0.60

»Ĵ

-0.60

ady

-0.59

POSITIVE LOGITS

 lows

0.89

 highs

0.84

Ago

0.77

 ILCS

0.71

nces

0.71

 consuming

0.66

asured

0.65

 favourite

0.63

Hig

0.63

orie

0.62

Activations Density 0.035%

references to all-time highs and lows

references to "all-time highs" in various contexts

No Comments

No Known Activations