LED's have a VERY NARROW linear operation curve for light output.
Red and Amber provide the better-wider knee of operation.
Green onto Blue - not so much. Because of the region or thickness of the substrate that emits their light - the VALENCE band - is very narrow. You only have a short distance (In Angstroms) in which the Electron bounces from one shell to another - the release energy region is far narrower than the Red or Ambers.
So what to do?
Well, there is a "divider" scheme that may help you determine the amount of current the LED can use to light with, and the rest just can pass on by.
So you already have the resistor - best to use one or pair of, with one in series with - a low value potentiometer to help you make a shunt - the energy the mod light lamp used before was in microamps (mA) so even then it didn't need much. But used VOLTAGE to offset this current to light the filament - that surge is what lit the lamp - in LED, you have plenty of power - the lamps' Silicon-based linear region is the problem.
LED's operate in the mA range too, but only drop so much voltage - their voltage drop - is like a step - one level - BAM - then another - then it TRIPs'- UP - to go up to another level then BAM down again, very little transition range to work with. It's a DIODE - with valve action - just no way to stop it down very well.
You'll need to use a resistor to stop down the power - but then place a resistor/potentiometer across the LED itself - to adjust it's window to pass power alongside the LED - you use a resistor/pot combo to force the dropping resistor to push power thru the LED, and not damage it from excessive power and current.
But to make it linear, you then have to shunt some of the power around the LED - so the Peaks and valleys of this modulation can be seen - that requires the shunt to bypass those peak power and low power changes more readily - else the LED sees all of it and tries to stay lit. The pot comes into play to help you find the linear region of where the voltage drop of the LED only takes in peaks - not everything else.
Some ideas have tried Diodes along side the LED - that would work, but limits the function to an on-off mode than to use it as a bright to dim linear region.
It's why I recommend the resistor-pot design.
Once the region (or required resistance) is found, that makes the LED work like you want it to, then the pot can be measured and a resistor can be put in it's place - or a series of resistors can be used to obtain that exact region. Across the LED Anode To Cathode - to offset the power passing thru the LED otherwise.
...runna pile-up on 6m SSB