Bespoke-Nimble-9B-vllm-cpp / temperature_config.json

Add Bespoke-Nimble-9B merged onto Qwen3.5-9B for vllm.cpp (convert-nimble.py)

52eead2 verified 22 days ago

294 Bytes

{
  "temperature": 1.0,
  "temperature_fitted": false,
  "adapter_sha256": "29ef39b072dee97287947455337879c1e916705c2f727287922a2d81f5e2f20a",
  "method": "softmax(candidate_logits / temperature)",
  "note": "Checkpoint-specific evaluated default; prior releases calibration does not apply."
}

	{
	"temperature": 1.0,
	"temperature_fitted": false,
	"adapter_sha256": "29ef39b072dee97287947455337879c1e916705c2f727287922a2d81f5e2f20a",
	"method": "softmax(candidate_logits / temperature)",
	"note": "Checkpoint-specific evaluated default; prior releases calibration does not apply."
	}