INDEX

Explanations

attempts or intentions to perform various actions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "attempted," indicating efforts or attempts made 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 @ 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

hander

-0.80

minus

-0.75

spr

-0.75

houses

-0.73

front

-0.70

sheet

-0.69

eyes

-0.69

today

-0.67

lined

-0.67

sheets

-0.66

POSITIVE LOGITS

 unsuccessfully

1.04

Attempts

0.84

ossibility

0.83

querque

0.80

 resusc

0.78

 attempts

0.78

TAIN

0.76

URES

0.76

llor

0.75

ossible

0.75

Activations Density 0.028%

attempts or intentions to perform various actions

instances of the word "attempted," indicating efforts or attempts made in various contexts

No Comments

No Known Activations