minor version update + major edits

[frgmt0]

added:

• beamWidth control to the tool input schema (beam_search)
• numSimulations control to the tool input schema (mcts)

[Jacck]

I've found a potential issue while testing: the numSimulations parameter is accepted but the MCTS strategy doesn't seem to reflect the changed value in its metrics (still shows default 50 simulations). Could you check if the simulation count is actually being applied in the MCTS implementation?

Otherwise the implementation looks clean with good input validation.

Regarding benchmark, lets have a talk, I share my email: [email protected]

[frgmt0]

That is definitely weird. I checked right now and here is the input output:

Default settings:

{
  `thought`: `Testing MCTS with default simulations`,
  `strategyType`: `mcts`,
  `thoughtNumber`: 1,
  `totalThoughts`: 3,
  `nextThoughtNeeded`: true
}
{"thoughtNumber":1,"totalThoughts":3,"nextThoughtNeeded":true,"thought":"Testing MCTS with default simulations","nodeId":"660b90de-bd83-46f0-9203-a810f0f5fd99","score":0.395,"strategyUsed":"mcts","stats":{"totalNodes":138,"averageScore":0.4418236714975843,"maxDepth":5,"branchingFactor":0.5434782608695652,"strategyMetrics":{"beam_search":{"name":"BeamSearchStrategy","nodesExplored":138,"averageScore":0.4418236714975843,"maxDepth":5,"beamWidth":3,"activeBeams":2,"totalBeamNodes":3},"mcts":{"name":"MonteCarloTreeSearchStrategy","nodesExplored":138,"averageScore":0.4418236714975843,"maxDepth":5,"simulationDepth":5,"numSimulations":50,"explorationConstant":1.4142135623730951,"totalSimulations":**51**,"active":true}}}}

I believe the default is 50 but im not sure why it lands on 51 unless the initial thought/start of the reasoning chain counts as one, which I think is the case

With 100:

{
  `thought`: `Testing MCTS with 100 simulations`,
  `strategyType`: `mcts`,
  `thoughtNumber`: 1,
  `totalThoughts`: 3,
  `numSimulations`: 100,
  `nextThoughtNeeded`: true
}
{"thoughtNumber":1,"totalThoughts":3,"nextThoughtNeeded":true,"thought":"Testing MCTS with 100 simulations","nodeId":"eec37abc-d605-4a9c-9aeb-844123e7b027","score":0.38833333333333336,"strategyUsed":"mcts","stats":{"totalNodes":144,"averageScore":0.4404861111111108,"maxDepth":5,"branchingFactor":0.5555555555555556,"strategyMetrics":{"beam_search":{"name":"BeamSearchStrategy","nodesExplored":144,"averageScore":0.4404861111111108,"maxDepth":5,"beamWidth":3,"activeBeams":2,"totalBeamNodes":3},"mcts":{"name":"MonteCarloTreeSearchStrategy","nodesExplored":144,"averageScore":0.4404861111111108,"maxDepth":5,"simulationDepth":5,"numSimulations":100,"explorationConstant":1.4142135623730951,"totalSimulations":**101**,"active":true}}}}

I wonder if the issue is that it's not reflected in the input?

As far as chatting you can reach me @ [email protected]

[frgmt0]

this one is more in depth and a major update id say

I also added some more verification for the mcts strategy that its being used and cleaned up some unused files.

I think for ToT (Tree of Thought) we can add that as a strategy later rather than a separate engine