PEFT compatible GEMM (#324)

Author: casper-hansen

awq/modules/linear/gemm.py

@@ -1,5 +1,6 @@
 import torch
 import torch.nn as nn
+from torch.autograd import Function
 from awq.utils.utils import get_best_device
 from awq.utils.packing_utils import dequantize_gemm
 
@@ -10,9 +11,94 @@
 except:
     AWQ_INSTALLED = False
 
+# Adapted from https://github.com/compressa-ai/AutoAWQ/tree/dev
+class WQLinearMMFunction(Function):
+    @staticmethod
+    # ctx is the first argument to forward
+    def forward(
+        ctx,
+        x,
+        qweight,
+        qzeros,
+        scales,
+        w_bit=4,
+        group_size=128,
+        bias=None,
+        out_features=0
+    ):
+        # The forward pass can use ctx.
+        ctx.save_for_backward(x, qweight, qzeros, scales, bias)
+        ctx.out_features = out_features
+
+        out_shape = x.shape[:-1] + (out_features, )
+        x = x.to(torch.float16)
+
+        if AWQ_INSTALLED:
+            FP16_MATMUL_HEURISTIC_CONDITION = x.shape[0]*x.shape[1] >= 1024
+
+            if FP16_MATMUL_HEURISTIC_CONDITION:
+                out = awq_ext.dequantize_weights_cuda(
+                    qweight,
+                    scales,
+                    qzeros,
+                    0,
+                    0,
+                    0,
+                    False
+                )
+                out = torch.matmul(x, out)
+            else:
+                out = awq_ext.gemm_forward_cuda(
+                    x.reshape(-1, x.shape[-1]),
+                    qweight,
+                    scales,
+                    qzeros,
+                    8
+                )
+        else:
+            out = dequantize_gemm(
+                qweight,
+                qzeros,
+                scales,
+                w_bit,
+                group_size
+            )
+            out = torch.matmul(x, out)
+
+        out = out + bias if bias is not None else out
+        out = out.reshape(out_shape)
+
+        # always want 3D tensor if tensor is 2D
+        if len(out.shape) == 2:
+            out = out.unsqueeze(0)
+        
+        return out
+
+    @staticmethod
+    def backward(ctx, grad_output):
+        input, qweight, qzeros, scales, bias = ctx.saved_tensors
+
+        weights = awq_ext.dequantize_weights_cuda(
+            qweight,
+            scales,
+            qzeros,
+            1,
+            0,
+            0,
+            False
+        )
+
+        if ctx.needs_input_grad[0]:
+            # 2D matrix multiplication, unsqueeze to 3D
+            grad_input = grad_output.squeeze(0).mm(
+                weights.transpose(0, 1)
+            ).unsqueeze(0)
+
+        return grad_input, None, None, None, None, None, None, None
+
 
 class WQLinear_GEMM(nn.Module):
-    def __init__(self, w_bit, group_size, in_features, out_features, bias, dev):
+    def __init__(self, w_bit, group_size, in_features, out_features, bias, dev, training=False):
         super().__init__()
 
         if w_bit not in [4]:
@@ -22,6 +108,7 @@ def __init__(self, w_bit, group_size, in_features, out_features, bias, dev):
         self.out_features = out_features
         self.w_bit = w_bit
         self.group_size = group_size if group_size != -1 else in_features
+        self.training = training
 
         # quick sanity check (make sure aligment)
         assert self.in_features % self.group_size == 0
@@ -145,45 +232,36 @@ def from_linear(
 
         return awq_linear
 
-    @torch.no_grad()
     def forward(self, x):
         out_shape = x.shape[:-1] + (self.out_features,)
 
         input_dtype = x.dtype
         if input_dtype != torch.float16:
             x = x.half()
 
-        if AWQ_INSTALLED:
-            FP16_MATMUL_HEURISTIC_CONDITION = x.shape[0] * x.shape[1] >= 1024
-
-            if FP16_MATMUL_HEURISTIC_CONDITION:
-                out = awq_ext.dequantize_weights_cuda(
-                    self.qweight,
-                    self.scales,
-                    self.qzeros,
-                    0,
-                    0,
-                    0,
-                    False,
-                )
-                out = torch.matmul(x, out)
-            else:
-                out = awq_ext.gemm_forward_cuda(
-                    x.reshape(-1, x.shape[-1]),
-                    self.qweight,
-                    self.scales,
-                    self.qzeros,
-                    8,
-                )
-        else:
-            out = dequantize_gemm(
+        if self.training:
+            out = WQLinearMMFunction.apply(
+                x,
                 self.qweight,
                 self.qzeros,
                 self.scales,
                 self.w_bit,
                 self.group_size,
+                self.bias,
+                self.out_features,
             )
-            out = torch.matmul(x, out)
+        else:
+            with torch.no_grad():
+                out = WQLinearMMFunction.apply(
+                    x,
+                    self.qweight,
+                    self.qzeros,
+                    self.scales,
+                    self.w_bit,
+                    self.group_size,
+                    self.bias,
+                    self.out_features,
+                )
 
         if input_dtype != torch.float16:
             out = out.to(dtype=input_dtype)

examples/awq_train.py

@@ -0,0 +1,76 @@
+import datasets
+from awq import AutoAWQForCausalLM
+from transformers import (
+    AutoTokenizer,
+    TrainingArguments,
+    Trainer,
+    DataCollatorForLanguageModeling
+)
+from peft import get_peft_model, LoraConfig, TaskType
+
+def prepare_split(tokenizer):
+    data = datasets.load_dataset("mhenrichsen/alpaca_2k_test", split="train")
+    prompt_template = "<s>[INST] {system} {prompt} [/INST] {output}</s>"
+
+    def format_prompt(x):
+        return prompt_template.format(
+            system="",
+            prompt=x["instruction"],
+            output=x["output"]
+        )
+
+    data = data.map(
+        lambda x: {"text": format_prompt(x)},
+    ).select_columns(["text"])
+    data = data.map(lambda x: tokenizer(x["text"]), batched=True)
+
+    return data
+
+model_path = "ybelkada/opt-125m-awq"
+
+# Load model
+model = AutoAWQForCausalLM.from_quantized(model_path, fuse_layers=False)
+tokenizer = AutoTokenizer.from_pretrained(model_path)
+tokenizer.pad_token = tokenizer.eos_token
+
+# Prepare data
+data_train = prepare_split(tokenizer)
+
+# Config Lora
+lora_config = LoraConfig(
+    r=4,
+    lora_alpha=8,
+    lora_dropout=0.5,
+    bias="none",
+    task_type=TaskType.CAUSAL_LM,
+    inference_mode=False
+)
+
+model = get_peft_model(model.model, lora_config)
+
+model.print_trainable_parameters()
+
+training_arguments = TrainingArguments(
+    output_dir="./output",
+    per_device_train_batch_size=1,
+    optim="adamw_torch",
+    num_train_epochs=1,
+    learning_rate=1e-4,
+    # fp16=True,
+    evaluation_strategy="no",
+    save_strategy="epoch",
+    save_steps=100,
+    logging_steps=50,
+    eval_steps=None,
+    load_best_model_at_end=False
+)
+
+trainer = Trainer(
+    model=model,
+    train_dataset=data_train,
+    args=training_arguments,
+    data_collator=DataCollatorForLanguageModeling(tokenizer, mlm=False),
+)
+
+trainer.train()
+trainer.save_model("output")