INDEX

Explanations

terms related to time durations, specifically highlighting events happening at the end of a specified time period

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "Last" indicating time references

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 parap

-0.66

 lying

-0.65

rig

-0.65

 steel

-0.63

 plant

-0.62

 discern

-0.62

 endeav

-0.61

 glass

-0.60

 inclined

-0.60

 wound

-0.60

POSITIVE LOGITS

Last

3.53

 Last

2.34

last

2.11

Yesterday

1.98

 LAST

1.93

Earlier

1.72

Recently

1.70

 last

1.53

Latest

1.50

Recent

1.49

Activations Density 0.010%

terms related to time durations, specifically highlighting events happening at the end of a specified time period

occurrences of the word "Last" indicating time references

No Comments

No Known Activations