liger-kernel-nightly 0.4.1.dev20241114155849__tar.gz → 0.4.1.dev20241115012952__tar.gz

{liger_kernel_nightly-0.4.1.dev20241114155849/src/liger_kernel_nightly.egg-info → liger_kernel_nightly-0.4.1.dev20241115012952}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: liger_kernel_nightly
-Version: 0.4.1.dev20241114155849
+Version: 0.4.1.dev20241115012952
 Summary: Efficient Triton kernels for LLM Training
 License: BSD 2-CLAUSE LICENSE
         Copyright 2024 LinkedIn Corporation 

{liger_kernel_nightly-0.4.1.dev20241114155849 → liger_kernel_nightly-0.4.1.dev20241115012952}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "liger_kernel_nightly"
-version = "0.4.1.dev20241114155849"
+version = "0.4.1.dev20241115012952"
 description = "Efficient Triton kernels for LLM Training"
 urls = { "Homepage" = "https://github.com/linkedin/Liger-Kernel" }
 readme = { file = "README.md", content-type = "text/markdown" }

liger_kernel_nightly-0.4.1.dev20241115012952/src/liger_kernel/chunked_loss/dpo_loss.py

@@ -0,0 +1,57 @@
+import torch.nn.functional as F
+
+from liger_kernel.chunked_loss.fused_linear_preference import (
+    LigerFusedLinearPreferenceBase,
+)
+
+
+class LigerFusedLinearDPOFunction(LigerFusedLinearPreferenceBase):
+
+    @staticmethod
+    def preference_loss_fn(chosen_logps, rejected_logps, beta=0.1):
+        """
+        Compute DPO loss (Direct Preference Optimization).
+        Args:
+            chosen_logps (torch.Tensor): Avg log probabilities of chosen tokens. Shape: (batch_size,).
+            rejected_logps (torch.Tensor): Avg log probabilities of rejected tokens. Shape: (batch_size,).
+            beta (float): Weight for the direct preference loss.
+        """
+        logits_diff = beta * (chosen_logps - rejected_logps)
+        losses = -F.logsigmoid(logits_diff)
+        return losses.sum()
+
+    @staticmethod
+    def forward(
+        ctx,
+        _input,
+        weight,
+        target,
+        bias=None,
+        ignore_index=-100,
+        beta=0.1,
+        compute_nll_loss=True,
+        compiled=True,
+    ):
+        """
+        Fused linear layer with DPO (Direct Preference Optimization) loss.
+        Handles both the forward and backward pass of the final linear layer with DPO loss.
+        """
+        return LigerFusedLinearPreferenceBase.forward(
+            ctx=ctx,
+            _input=_input,
+            weight=weight,
+            target=target,
+            bias=bias,
+            loss_fn=LigerFusedLinearDPOFunction.preference_loss_fn,
+            compute_nll_loss=compute_nll_loss,
+            ignore_index=ignore_index,
+            beta=beta,
+            compiled=compiled,
+        )
+
+    @staticmethod
+    def backward(ctx, grad_output):
+        # Get gradients for _input, weight, bias, and target from the base class
+        grads = LigerFusedLinearPreferenceBase.backward(ctx, grad_output)[:4]
+        # Return these gradients, followed by None for the remaining inputs
+        return *grads, None, None, None, None

{liger_kernel_nightly-0.4.1.dev20241114155849 → liger_kernel_nightly-0.4.1.dev20241115012952/src/liger_kernel_nightly.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: liger_kernel_nightly
-Version: 0.4.1.dev20241114155849
+Version: 0.4.1.dev20241115012952
 Summary: Efficient Triton kernels for LLM Training
 License: BSD 2-CLAUSE LICENSE
         Copyright 2024 LinkedIn Corporation 

{liger_kernel_nightly-0.4.1.dev20241114155849 → liger_kernel_nightly-0.4.1.dev20241115012952}/src/liger_kernel_nightly.egg-info/SOURCES.txt

@@ -4,6 +4,7 @@ README.md
 pyproject.toml
 src/liger_kernel/env_report.py
 src/liger_kernel/chunked_loss/__init__.py
+src/liger_kernel/chunked_loss/dpo_loss.py
 src/liger_kernel/chunked_loss/fused_linear_preference.py
 src/liger_kernel/chunked_loss/orpo_loss.py
 src/liger_kernel/ops/__init__.py