added inference defaults from unsloth guides

studio/backend/assets/configs/model_defaults/gemma/unsloth_functiongemma-270m-it.yaml

@@ -1,6 +1,7 @@
 # Model defaults for unsloth/functiongemma-270m-it
 # Based on FunctionGemma_(270M).ipynb
 # Also applies to: unsloth/functiongemma-270m-it-unsloth-bnb-4bit, google/functiongemma-270m-it, unsloth/functiongemma-270m-it-unsloth-bnb-4bit
+# added inference parameters from unsloth guides
 
 training:
   max_seq_length: 4096
@@ -45,3 +46,9 @@ logging:
   tensorboard_dir: "runs"
   log_frequency: 10
 
+inference:
+  temperature: 1.0
+  top_k: 64
+  top_p: 0.95
+  min_p: 0.0
+

studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-3-270m-it.yaml

@@ -1,6 +1,7 @@
 # Model defaults for unsloth/gemma-3-270m-it
 # Based on Gemma3_(270M).ipynb
 # Also applies to: unsloth/gemma-3-270m-it-unsloth-bnb-4bit, google/gemma-3-270m-it, unsloth/gemma-3-270m-it-bnb-4bit
+# added inference parameters from unsloth guides
 
 training:
   max_seq_length: 2048
@@ -45,3 +46,9 @@ logging:
   tensorboard_dir: "runs"
   log_frequency: 10
 
+inference:
+  temperature: 1.0
+  top_k: 64
+  top_p: 0.95
+  min_p: 0.0
+

studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-3-27b-it.yaml

@@ -1,6 +1,7 @@
 # Model defaults for unsloth/gemma-3-27b-it
 # Based on Gemma3_(27B)_A100-Conversational.ipynb
 # Also applies to: unsloth/gemma-3-27b-it-unsloth-bnb-4bit, google/gemma-3-27b-it, unsloth/gemma-3-27b-it-bnb-4bit
+# added inference parameters from unsloth guides
 
 training:
   max_seq_length: 2048
@@ -39,3 +40,9 @@ logging:
   tensorboard_dir: "runs"
   log_frequency: 10
 
+inference:
+  temperature: 1.0
+  top_k: 64
+  top_p: 0.95
+  min_p: 0.0
+

studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-3-4b-it.yaml

@@ -1,6 +1,7 @@
 # Model defaults for unsloth/gemma-3-4b-it
 # Based on Gemma3_(4B).ipynb
 # Also applies to: unsloth/gemma-3-4b-it-unsloth-bnb-4bit, google/gemma-3-4b-it, unsloth/gemma-3-4b-it-bnb-4bit
+# added inference parameters from unsloth guides
 
 training:
   max_seq_length: 2048
@@ -39,3 +40,9 @@ logging:
   tensorboard_dir: "runs"
   log_frequency: 10
 
+inference:
+  temperature: 1.0
+  top_k: 64
+  top_p: 0.95
+  min_p: 0.0
+

studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-3-4b-pt.yaml

@@ -1,6 +1,7 @@
 # Model defaults for unsloth/gemma-3-4b-pt
 # Based on Gemma3_(4B)-Vision.ipynb
 # Also applies to: unsloth/gemma-3-4b-pt-unsloth-bnb-4bit, google/gemma-3-4b-pt, unsloth/gemma-3-4b-pt-bnb-4bit
+# added inference parameters from unsloth guides
 
 training:
   max_seq_length: 2048
@@ -39,3 +40,9 @@ logging:
   tensorboard_dir: "runs"
   log_frequency: 10
 
+inference:
+  temperature: 1.0
+  top_k: 64
+  top_p: 0.95
+  min_p: 0.0
+

studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-3n-E4B-it.yaml

@@ -1,6 +1,7 @@
 # Model defaults for unsloth/gemma-3n-E4B-it
 # Based on Gemma3N_(4B)-Conversational.ipynb
 # Also applies to: unsloth/gemma-3n-E4B-it-unsloth-bnb-4bit, google/gemma-3n-E4B-it, unsloth/gemma-3n-E4B-it-unsloth-bnb-4bit
+# added inference parameters from unsloth guides
 
 training:
   max_seq_length: 1024
@@ -39,3 +40,9 @@ logging:
   tensorboard_dir: "runs"
   log_frequency: 10
 
+inference:
+  temperature: 1.0
+  top_k: 64
+  top_p: 0.95
+  min_p: 0.0
+

studio/backend/assets/configs/model_defaults/gemma/unsloth_gemma-3n-E4B.yaml

@@ -1,6 +1,7 @@
 # Model defaults for unsloth/gemma-3n-E4B
 # Based on Gemma3N_(4B)-Vision.ipynb
 # Also applies to: unsloth/gemma-3n-E4B-unsloth-bnb-4bit, google/gemma-3n-E4B
+# added inference parameters from unsloth guides
 
 training:
   max_seq_length: 2048
@@ -39,3 +40,9 @@ logging:
   tensorboard_dir: "runs"
   log_frequency: 10
 
+inference:
+  temperature: 1.0
+  top_k: 64
+  top_p: 0.95
+  min_p: 0.0
+

studio/backend/assets/configs/model_defaults/gpt-oss/unsloth_gpt-oss-120b.yaml

@@ -1,6 +1,7 @@
 # Model defaults for unsloth/gpt-oss-120b
 # Based on gpt-oss-(120B)_A100-Fine-tuning.ipynb
 # Also applies to: openai/gpt-oss-120b, unsloth/gpt-oss-120b-unsloth-bnb-4bit
+# added inference parameters from unsloth guides
 
 training:
   max_seq_length: 4096
@@ -45,3 +46,8 @@ logging:
   tensorboard_dir: "runs"
   log_frequency: 10
 
+inference:
+  temperature: 1.0
+  top_p: 1.0
+  top_k: 0
+

studio/backend/assets/configs/model_defaults/gpt-oss/unsloth_gpt-oss-20b.yaml

@@ -1,6 +1,7 @@
 # Model defaults for unsloth/gpt-oss-20b
 # Based on gpt-oss-(20B)-Fine-tuning.ipynb
 # Also applies to: openai/gpt-oss-20b, unsloth/gpt-oss-20b-unsloth-bnb-4bit, unsloth/gpt-oss-20b-BF16
+# added inference parameters from unsloth guides
 
 training:
   max_seq_length: 1024
@@ -45,3 +46,8 @@ logging:
   tensorboard_dir: "runs"
   log_frequency: 10
 
+inference:
+  temperature: 1.0
+  top_p: 1.0
+  top_k: 0
+

studio/backend/assets/configs/model_defaults/granite/unsloth_granite-4.0-350m-unsloth-bnb-4bit.yaml

@@ -1,6 +1,7 @@
 # Model defaults for unsloth/granite-4.0-350m
 # Based on Granite4.0_350M.ipynb
 # Also applies to: ibm-granite/granite-4.0-350m, unsloth/granite-4.0-350m-bnb-4bit
+# added inference parameters from unsloth guides
 
 training:
   max_seq_length: 2048
@@ -47,3 +48,8 @@ logging:
   tensorboard_dir: "runs"
   log_frequency: 10
 
+inference:
+  temperature: 0.0
+  top_p: 1.0
+  top_k: 0
+

studio/backend/assets/configs/model_defaults/granite/unsloth_granite-4.0-h-micro.yaml

@@ -1,6 +1,7 @@
 # Model defaults for unsloth/granite-4.0-h-micro
 # Based on Granite4.0.ipynb
 # Also applies to: ibm-granite/granite-4.0-h-micro, unsloth/granite-4.0-h-micro-bnb-4bit, unsloth/granite-4.0-h-micro-unsloth-bnb-4bit
+# added inference parameters from unsloth guides
 
 training:
   max_seq_length: 2048
@@ -47,3 +48,8 @@ logging:
   tensorboard_dir: "runs"
   log_frequency: 10
 
+inference:
+  temperature: 0.0
+  top_p: 1.0
+  top_k: 0
+

studio/backend/assets/configs/model_defaults/mistral/unsloth_Magistral-Small-2509-unsloth-bnb-4bit.yaml

@@ -1,6 +1,7 @@
 # Model defaults for unsloth/Magistral-Small-2509
 # Based on Magistral_(24B)-Reasoning-Conversational.ipynb
 # Also applies to: mistralai/Magistral-Small-2509, unsloth/Magistral-Small-2509-bnb-4bit
+# added inference parameters from unsloth guides
 
 training:
   max_seq_length: 2048
@@ -45,3 +46,8 @@ logging:
   tensorboard_dir: "runs"
   log_frequency: 10
 
+inference:
+  temperature: 0.7
+  min_p: 0.01
+  top_p: 0.95
+

studio/backend/assets/configs/model_defaults/mistral/unsloth_Ministral-3-3B-Instruct-2512.yaml

@@ -1,6 +1,7 @@
 # Model defaults for unsloth/Ministral-3-3B-Instruct-2512
 # Based on Ministral_3_VL_(3B)_Vision.ipynb
 # Also applies to: unsloth/Ministral-3-3B-Instruct-2512
+# added inference parameters from unsloth guides
 
 training:
   max_seq_length: 2048
@@ -45,3 +46,7 @@ logging:
   tensorboard_dir: "runs"
   log_frequency: 10
 
+inference:
+  temperature: 0.15
+  top_p: default
+

studio/backend/assets/configs/model_defaults/other/unsloth_Nemotron-3-Nano-30B-A3B.yaml

@@ -1,6 +1,7 @@
 # Model defaults for unsloth/Nemotron-3-Nano-30B-A3B
 # Based on Nemotron-3-Nano-30B-A3B_A100.ipynb
 # Also applies to: unsloth/Nemotron-3-Nano-30B-A3B
+# added inference parameters from unsloth guides
 
 training:
   max_seq_length: 2048
@@ -47,3 +48,7 @@ logging:
   tensorboard_dir: "runs"
   log_frequency: 10
 
+inference:
+  temperature: 1.0
+  top_p: 1.0
+

studio/backend/assets/configs/model_defaults/phi/unsloth_Phi-4.yaml

@@ -1,6 +1,7 @@
 # Model defaults for unsloth/Phi-4
 # Based on Phi_4-Conversational.ipynb
 # Also applies to: unsloth/phi-4-unsloth-bnb-4bit, microsoft/phi-4, unsloth/phi-4-bnb-4bit
+# added inference parameters from unsloth guides
 
 training:
   max_seq_length: 2048
@@ -45,3 +46,7 @@ logging:
   tensorboard_dir: "runs"
   log_frequency: 10
 
+inference:
+  temperature: 0.8
+  top_p: 0.95
+

studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen3-4B-Instruct-2507.yaml

@@ -1,6 +1,7 @@
 # Model defaults for unsloth/Qwen3-4B-Instruct-2507
 # Based on Qwen3_(4B)-Instruct.ipynb
 # Also applies to: unsloth/Qwen3-4B-Instruct-2507-unsloth-bnb-4bit, Qwen/Qwen3-4B-Instruct-2507, unsloth/Qwen3-4B-Instruct-2507-bnb-4bit, Qwen/Qwen3-4B-Instruct-2507-FP8, unsloth/Qwen3-4B-Instruct-2507-FP8
+# added inference parameters from unsloth guides
 
 training:
   max_seq_length: 2048
@@ -45,3 +46,9 @@ logging:
   tensorboard_dir: "runs"
   log_frequency: 10
 
+inference:
+  temperature: 0.7
+  top_p: 0.80
+  top_k: 20
+  min_p: 0.00
+

studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen3-4B-Thinking-2507.yaml

@@ -1,6 +1,7 @@
 # Model defaults for unsloth/Qwen3-4B-Thinking-2507
 # Based on Qwen3_(4B)-Thinking.ipynb
 # Also applies to: unsloth/Qwen3-4B-Thinking-2507-unsloth-bnb-4bit, Qwen/Qwen3-4B-Thinking-2507, unsloth/Qwen3-4B-Thinking-2507-bnb-4bit, Qwen/Qwen3-4B-Thinking-2507-FP8, unsloth/Qwen3-4B-Thinking-2507-FP8
+# added inference parameters from unsloth guides
 
 training:
   max_seq_length: 2048
@@ -45,3 +46,9 @@ logging:
   tensorboard_dir: "runs"
   log_frequency: 10
 
+inference:
+  temperature: 0.6
+  top_p: 0.95
+  top_k: 20
+  min_p: 0.00
+

studio/backend/assets/configs/model_defaults/qwen/unsloth_Qwen3-VL-8B-Instruct-unsloth-bnb-4bit.yaml

@@ -1,6 +1,7 @@
 # Model defaults for unsloth/Qwen3-VL-8B-Instruct
 # Based on Qwen3_VL_(8B)-Vision.ipynb
 # Also applies to: Qwen/Qwen3-VL-8B-Instruct-FP8, unsloth/Qwen3-VL-8B-Instruct-FP8, unsloth/Qwen3-VL-8B-Instruct, Qwen/Qwen3-VL-8B-Instruct, unsloth/Qwen3-VL-8B-Instruct-bnb-4bit
+# added inference parameters from unsloth guides
 
 training:
   max_seq_length: 2048
@@ -39,3 +40,8 @@ logging:
   tensorboard_dir: "runs"
   log_frequency: 10
 
+inference:
+  temperature: 0.7
+  top_p: 0.8
+  top_k: 20
+