remove dup quant parsing from individual model class

YouNeedCryDear · YouNeedCryDear · commit a1678258fe80 · 2026-04-01T12:12:26.000-07:00
diff --git a/pkg/hfutil/modelconfig/baichuan.go b/pkg/hfutil/modelconfig/baichuan.go
@@ -88,11 +88,6 @@ func (c *BaichuanConfig) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), c.TorchDtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *BaichuanConfig) GetQuantizationType() string {
-	return "" // No quantization by default
-}
-
 // HasVision returns false for Baichuan base models
 func (c *BaichuanConfig) HasVision() bool {
 	return false
diff --git a/pkg/hfutil/modelconfig/bert.go b/pkg/hfutil/modelconfig/bert.go
@@ -109,11 +109,6 @@ func (c *BertConfig) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), c.TorchDtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *BertConfig) GetQuantizationType() string {
-	return "" // No quantization by default
-}
-
 // HasVision returns false for BERT models
 func (c *BertConfig) HasVision() bool {
 	return false
diff --git a/pkg/hfutil/modelconfig/chatglm.go b/pkg/hfutil/modelconfig/chatglm.go
@@ -111,11 +111,6 @@ func (c *ChatGLMConfig) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), dtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *ChatGLMConfig) GetQuantizationType() string {
-	return "" // No quantization by default
-}
-
 // HasVision returns false for ChatGLM base models
 func (c *ChatGLMConfig) HasVision() bool {
 	return false
diff --git a/pkg/hfutil/modelconfig/command_r.go b/pkg/hfutil/modelconfig/command_r.go
@@ -100,11 +100,6 @@ func (c *CommandRConfig) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), c.TorchDtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *CommandRConfig) GetQuantizationType() string {
-	return "" // No quantization by default
-}
-
 // HasVision returns false for Command-R base models
 func (c *CommandRConfig) HasVision() bool {
 	return false
diff --git a/pkg/hfutil/modelconfig/dbrx.go b/pkg/hfutil/modelconfig/dbrx.go
@@ -130,11 +130,6 @@ func (c *DBRXConfig) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), c.TorchDtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *DBRXConfig) GetQuantizationType() string {
-	return "" // No quantization by default
-}
-
 // HasVision returns false for DBRX base models
 func (c *DBRXConfig) HasVision() bool {
 	return false
diff --git a/pkg/hfutil/modelconfig/deepseek_v3.go b/pkg/hfutil/modelconfig/deepseek_v3.go
@@ -64,9 +64,6 @@ type DeepseekV3Config struct {
 	// RoPE scaling
 	RopeScaling RopeScalingConfig `json:"rope_scaling"`
 
-	// Quantization settings
-	QuantizationConfig *QuantizationConfig `json:"quantization_config,omitempty"`
-
 	// Misc options
 	TieWordEmbeddings bool    `json:"tie_word_embeddings"`
 	UseCache          bool    `json:"use_cache"`
@@ -146,14 +143,6 @@ func (c *DeepseekV3Config) GetTransformerVersion() string {
 	return c.BaseModelConfig.TransformerVersion
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *DeepseekV3Config) GetQuantizationType() string {
-	if c.QuantizationConfig != nil && c.QuantizationConfig.QuantMethod != "" {
-		return c.QuantizationConfig.QuantMethod
-	}
-	return ""
-}
-
 // GetArchitecture returns the model architecture
 func (c *DeepseekV3Config) GetArchitecture() string {
 	if len(c.Architectures) > 0 {
diff --git a/pkg/hfutil/modelconfig/deepseek_vl.go b/pkg/hfutil/modelconfig/deepseek_vl.go
@@ -171,11 +171,6 @@ func (c *DeepSeekVLConfig) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), dtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *DeepSeekVLConfig) GetQuantizationType() string {
-	return "" // No quantization by default
-}
-
 // HasVision returns true for DeepSeek VL models
 func (c *DeepSeekVLConfig) HasVision() bool {
 	return true
diff --git a/pkg/hfutil/modelconfig/exaone.go b/pkg/hfutil/modelconfig/exaone.go
@@ -102,11 +102,6 @@ func (c *ExaoneConfig) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), c.TorchDtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *ExaoneConfig) GetQuantizationType() string {
-	return "" // No quantization by default
-}
-
 // HasVision returns false for ExaONE base models
 func (c *ExaoneConfig) HasVision() bool {
 	return false
diff --git a/pkg/hfutil/modelconfig/gemma.go b/pkg/hfutil/modelconfig/gemma.go
@@ -127,11 +127,6 @@ func (c *GemmaConfig) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), c.TorchDtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *GemmaConfig) GetQuantizationType() string {
-	return "" // No quantization by default
-}
-
 // HasVision returns false for Gemma base models
 func (c *GemmaConfig) HasVision() bool {
 	return false
diff --git a/pkg/hfutil/modelconfig/gemma3.go b/pkg/hfutil/modelconfig/gemma3.go
@@ -133,11 +133,6 @@ func (c *Gemma3Config) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), c.TorchDtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *Gemma3Config) GetQuantizationType() string {
-	return "" // No quantization by default
-}
-
 // HasVision returns true since Gemma3 is a multimodal vision model
 func (c *Gemma3Config) HasVision() bool {
 	return true
diff --git a/pkg/hfutil/modelconfig/internlm.go b/pkg/hfutil/modelconfig/internlm.go
@@ -89,11 +89,6 @@ func (c *InternLMConfig) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), c.TorchDtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *InternLMConfig) GetQuantizationType() string {
-	return "" // No quantization by default
-}
-
 // HasVision returns false for InternLM base models
 func (c *InternLMConfig) HasVision() bool {
 	return false
diff --git a/pkg/hfutil/modelconfig/kimi_k2.go b/pkg/hfutil/modelconfig/kimi_k2.go
@@ -60,9 +60,6 @@ type KimiK2Config struct {
 	// RoPE scaling (YARN type for Kimi-K2)
 	RopeScaling RopeScalingConfig `json:"rope_scaling"`
 
-	// Quantization settings
-	QuantizationConfig *QuantizationConfig `json:"quantization_config,omitempty"`
-
 	// Misc options
 	TieWordEmbeddings bool    `json:"tie_word_embeddings"`
 	UseCache          bool    `json:"use_cache"`
@@ -141,14 +138,6 @@ func (c *KimiK2Config) GetTransformerVersion() string {
 	return c.BaseModelConfig.TransformerVersion
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *KimiK2Config) GetQuantizationType() string {
-	if c.QuantizationConfig != nil && c.QuantizationConfig.QuantMethod != "" {
-		return c.QuantizationConfig.QuantMethod
-	}
-	return ""
-}
-
 // GetArchitecture returns the model architecture
 func (c *KimiK2Config) GetArchitecture() string {
 	if len(c.Architectures) > 0 {
diff --git a/pkg/hfutil/modelconfig/llava.go b/pkg/hfutil/modelconfig/llava.go
@@ -157,11 +157,6 @@ func (c *LLaVAConfig) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), dtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *LLaVAConfig) GetQuantizationType() string {
-	return "" // No quantization by default
-}
-
 // HasVision returns true for LLaVA models
 func (c *LLaVAConfig) HasVision() bool {
 	return true
diff --git a/pkg/hfutil/modelconfig/minicpm.go b/pkg/hfutil/modelconfig/minicpm.go
@@ -95,11 +95,6 @@ func (c *MiniCPMConfig) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), c.TorchDtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *MiniCPMConfig) GetQuantizationType() string {
-	return "" // No quantization by default
-}
-
 // HasVision returns false for base MiniCPM models
 func (c *MiniCPMConfig) HasVision() bool {
 	return false
diff --git a/pkg/hfutil/modelconfig/mistral.go b/pkg/hfutil/modelconfig/mistral.go
@@ -108,11 +108,6 @@ func (c *MistralConfig) GetTransformerVersion() string {
 	return c.TransformerVersion
 }
 
-func (c *MistralConfig) GetQuantizationType() string {
-	// Mistral doesn't have quantization in this config
-	return ""
-}
-
 func (c *MistralConfig) GetArchitecture() string {
 	if len(c.Architectures) > 0 {
 		return c.Architectures[0]
diff --git a/pkg/hfutil/modelconfig/mixtral.go b/pkg/hfutil/modelconfig/mixtral.go
@@ -117,11 +117,6 @@ func (c *MixtralConfig) GetTransformerVersion() string {
 	return c.TransformerVersion
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *MixtralConfig) GetQuantizationType() string {
-	return "" // No quantization config for Mixtral by default
-}
-
 // GetArchitecture returns the model architecture
 func (c *MixtralConfig) GetArchitecture() string {
 	if len(c.Architectures) > 0 {
diff --git a/pkg/hfutil/modelconfig/mllama.go b/pkg/hfutil/modelconfig/mllama.go
@@ -145,11 +145,6 @@ func (c *MLlamaConfig) GetTransformerVersion() string {
 	return c.TransformerVersion
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *MLlamaConfig) GetQuantizationType() string {
-	return "" // MLlama models don't have quantization by default
-}
-
 // GetArchitecture returns the model architecture
 func (c *MLlamaConfig) GetArchitecture() string {
 	if len(c.Architectures) > 0 {
diff --git a/pkg/hfutil/modelconfig/phi3.go b/pkg/hfutil/modelconfig/phi3.go
@@ -103,11 +103,6 @@ func (c *Phi3Config) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), c.TorchDtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *Phi3Config) GetQuantizationType() string {
-	return "" // No quantization by default
-}
-
 // HasVision returns false since this is not a multimodal vision model
 func (c *Phi3Config) HasVision() bool {
 	return false
diff --git a/pkg/hfutil/modelconfig/phi3_v.go b/pkg/hfutil/modelconfig/phi3_v.go
@@ -92,11 +92,6 @@ func (c *Phi3VConfig) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), c.TorchDtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *Phi3VConfig) GetQuantizationType() string {
-	return "" // No quantization by default
-}
-
 // HasVision returns true since this is a multimodal vision model
 func (c *Phi3VConfig) HasVision() bool {
 	return c.ImgProcessor != nil
diff --git a/pkg/hfutil/modelconfig/phi3small.go b/pkg/hfutil/modelconfig/phi3small.go
@@ -114,11 +114,6 @@ func (c *Phi3SmallConfig) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), c.TorchDtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *Phi3SmallConfig) GetQuantizationType() string {
-	return "" // No quantization by default
-}
-
 // HasVision returns false since this is not a multimodal vision model
 func (c *Phi3SmallConfig) HasVision() bool {
 	return false
diff --git a/pkg/hfutil/modelconfig/phimoe.go b/pkg/hfutil/modelconfig/phimoe.go
@@ -116,11 +116,6 @@ func (c *PhiMoEConfig) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), c.TorchDtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *PhiMoEConfig) GetQuantizationType() string {
-	return "" // No quantization by default
-}
-
 // HasVision returns false since this is not a multimodal vision model
 func (c *PhiMoEConfig) HasVision() bool {
 	return false
diff --git a/pkg/hfutil/modelconfig/qwen.go b/pkg/hfutil/modelconfig/qwen.go
@@ -48,9 +48,6 @@ type QwenConfig struct {
 	FP32           bool   `json:"fp32"`
 	OnnxSafe       *bool  `json:"onnx_safe"`
 	TokenizerClass string `json:"tokenizer_class"`
-
-	// Quantization
-	QuantizationConfig *QuantizationConfig `json:"quantization_config,omitempty"`
 }
 
 // LoadQwenConfig loads a Qwen v1 model configuration from a JSON file
@@ -110,14 +107,6 @@ func (c *QwenConfig) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), c.TorchDtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *QwenConfig) GetQuantizationType() string {
-	if c.QuantizationConfig != nil && c.QuantizationConfig.QuantMethod != "" {
-		return c.QuantizationConfig.QuantMethod
-	}
-	return ""
-}
-
 // HasVision returns false for Qwen v1 base models
 func (c *QwenConfig) HasVision() bool {
 	return false // Base Qwen v1 models don't have vision capabilities
diff --git a/pkg/hfutil/modelconfig/qwen2.go b/pkg/hfutil/modelconfig/qwen2.go
@@ -39,9 +39,6 @@ type Qwen2Config struct {
 	TieWordEmbeddings bool `json:"tie_word_embeddings"`
 	UseCache          bool `json:"use_cache"`
 	UseSlidingWindow  bool `json:"use_sliding_window"`
-
-	// Quantization
-	QuantizationConfig *QuantizationConfig `json:"quantization_config,omitempty"`
 }
 
 // LoadQwen2Config loads a Qwen2 model configuration from a JSON file
@@ -100,14 +97,6 @@ func (c *Qwen2Config) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), c.TorchDtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *Qwen2Config) GetQuantizationType() string {
-	if c.QuantizationConfig != nil && c.QuantizationConfig.QuantMethod != "" {
-		return c.QuantizationConfig.QuantMethod
-	}
-	return ""
-}
-
 // HasVision returns false for Qwen2 base models
 func (c *Qwen2Config) HasVision() bool {
 	return false // Base Qwen2 models don't have vision capabilities
diff --git a/pkg/hfutil/modelconfig/qwen2_vl.go b/pkg/hfutil/modelconfig/qwen2_vl.go
@@ -75,9 +75,6 @@ type Qwen2VLConfig struct {
 	TieWordEmbeddings bool `json:"tie_word_embeddings"`
 	UseCache          bool `json:"use_cache"`
 	UseSlidingWindow  bool `json:"use_sliding_window"`
-
-	// Quantization
-	QuantizationConfig *QuantizationConfig `json:"quantization_config,omitempty"`
 }
 
 // LoadQwen2VLConfig loads a Qwen2-VL model configuration from a JSON file
@@ -141,14 +138,6 @@ func (c *Qwen2VLConfig) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), c.TorchDtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *Qwen2VLConfig) GetQuantizationType() string {
-	if c.QuantizationConfig != nil && c.QuantizationConfig.QuantMethod != "" {
-		return c.QuantizationConfig.QuantMethod
-	}
-	return ""
-}
-
 // HasVision returns true for Qwen2-VL models
 func (c *Qwen2VLConfig) HasVision() bool {
 	return true // Qwen2-VL models have vision capabilities
diff --git a/pkg/hfutil/modelconfig/qwen3.go b/pkg/hfutil/modelconfig/qwen3.go
@@ -45,9 +45,6 @@ type Qwen3Config struct {
 
 	// Embedding config
 	SimilarityFnName string `json:"similarity_fn_name"`
-
-	// Quantization
-	QuantizationConfig *QuantizationConfig `json:"quantization_config,omitempty"`
 }
 
 // LoadQwen3Config loads a Qwen3 model configuration from a JSON file
@@ -104,14 +101,6 @@ func (c *Qwen3Config) GetModelSizeBytes() int64 {
 	return EstimateModelSizeBytes(c.GetParameterCount(), c.TorchDtype)
 }
 
-// GetQuantizationType returns the quantization method used (if any)
-func (c *Qwen3Config) GetQuantizationType() string {
-	if c.QuantizationConfig != nil && c.QuantizationConfig.QuantMethod != "" {
-		return c.QuantizationConfig.QuantMethod
-	}
-	return ""
-}
-
 // HasVision returns false for Qwen3 base models
 func (c *Qwen3Config) HasVision() bool {
 	return false // Base Qwen3 models don't have vision capabilities
diff --git a/pkg/hfutil/modelconfig/qwen3_moe.go b/pkg/hfutil/modelconfig/qwen3_moe.go
diff --git a/pkg/hfutil/modelconfig/qwen3_vl.go b/pkg/hfutil/modelconfig/qwen3_vl.go
diff --git a/pkg/hfutil/modelconfig/stablelm.go b/pkg/hfutil/modelconfig/stablelm.go
diff --git a/pkg/hfutil/modelconfig/xverse.go b/pkg/hfutil/modelconfig/xverse.go
