Fix GRPO eval

pyproject.toml

@@ -37,7 +37,7 @@ triton = [
 ]
 
 huggingface = [
-    "unsloth_zoo>=2025.4.4",
+    "unsloth_zoo>=2025.5.1",
     "packaging",
     "tyro",
     "transformers>=4.51.3,!=4.47.0",
@@ -381,7 +381,7 @@ colab-ampere-torch220 = [
     "flash-attn>=2.6.3",
 ]
 colab-new = [
-    "unsloth_zoo>=2025.4.4",
+    "unsloth_zoo>=2025.5.1",
     "packaging",
     "tyro",
     "transformers>=4.51.3,!=4.47.0",

unsloth/models/_utils.py

@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-__version__ = "2025.4.8"
+__version__ = "2025.5.1"
 
 __all__ = [
     "SUPPORTS_BFLOAT16",

unsloth/models/llama.py

@@ -1076,7 +1076,7 @@ def CausalLM_fast_forward(fast_forward_inference):
         if labels is not None: labels = labels.to(lm_head_device)
 
         # Output last hidden states without logits if asked
-        if self.training and os.environ.get("UNSLOTH_RETURN_HIDDEN_STATES", "0") == "1":
+        if os.environ.get("UNSLOTH_RETURN_HIDDEN_STATES", "0") == "1":
             if num_logits_to_keep != 0:
                 hidden_states = hidden_states[:, -num_logits_to_keep:, :]
             return CausalLMOutputWithPast(