INDEX

Explanations

words related to completion, finalization, or achievement

oai_token-act-pair · gpt-3.5-turbo

phrases related to completing or finishing actions and events

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

Magikarp

-0.75

 Cosponsors

-0.71

ITNESS

-0.67

 counsel

-0.66

soDeliveryDate

-0.65

soType

-0.63

 affiliates

-0.62

Filename

-0.61

 pitches

-0.60

groups

-0.60

POSITIVE LOGITS

cellence

0.82

bye

0.74

Ĥ¬

0.72

teness

0.70

ocre

0.70

angering

0.70

icking

0.69

lished

0.68

 overc

0.68

 throat

0.67

Activations Density 0.079%

words related to completion, finalization, or achievement

phrases related to completing or finishing actions and events

No Comments

No Known Activations