Model: More Jax import fixes and formatting

modeling/inference_model.py

@@ -16,13 +16,6 @@ from modeling.tokenizer import GenericTokenizer
 
 import utils
 
-try:
-    import tpu_mtj_backend
-except ModuleNotFoundError as e:
-    # Not on TPU... hopefully
-    if utils.koboldai_vars.use_colab_tpu:
-        raise e
-
 # We only want to use logit manipulations and such on our core text model
 class use_core_manipulations:
     """Use in a `with` block to patch functions for core story model sampling."""

modeling/inference_models/hf_mtj.py

@@ -18,12 +18,9 @@ from modeling.inference_model import (
 )
 from modeling.inference_models.hf import HFInferenceModel
 
-try:
+# This file shouldn't be imported unless using the TPU
+assert utils.koboldai_vars.use_colab_tpu
 import tpu_mtj_backend
-except ModuleNotFoundError as e:
-    # Not on TPU... hopefully
-    if utils.koboldai_vars.use_colab_tpu:
-        raise e
 
 
 class HFMTJInferenceModel(HFInferenceModel):
@@ -39,7 +36,7 @@ class HFMTJInferenceModel(HFInferenceModel):
             post_token_hooks=False,
             stopper_hooks=False,
             post_token_probs=False,
-            uses_tpu=True
+            uses_tpu=True,
         )
 
     def setup_mtj(self) -> None:

modeling/inference_models/rwkv.py

@@ -217,10 +217,7 @@ class RWKVInferenceModel(InferenceModel):
             for _ in range(max_new):
 
                 logits, state = self.model.forward([last_token], state)
-                last_token = self._sample_token(
-                    logits,
-                    context
-                )
+                last_token = self._sample_token(logits, context)
                 out.append(last_token)
                 add = torch.tensor([[last_token]]).to(aux_device)
                 context = torch.cat((context, add), dim=-1)

modeling/test_generation.py

@@ -18,6 +18,7 @@ TEST_SEED = 1337
 
 # HF Torch
 
+
 def test_generic_hf_torch_load() -> None:
     global model
     model = GenericHFTorchInferenceModel(
@@ -38,7 +39,9 @@ def test_torch_inference() -> None:
     x = model.raw_generate(TEST_PROMPT, max_new=TEST_GEN_TOKEN_COUNT, seed=TEST_SEED)
     print(x.decoded)
     assert len(x.encoded) == 1, "Bad output shape (too many batches!)"
-    assert len(x.encoded[0]) == TEST_GEN_TOKEN_COUNT, f"Wrong token amount (requested {TEST_GEN_TOKEN_COUNT})"
+    assert (
+        len(x.encoded[0]) == TEST_GEN_TOKEN_COUNT
+    ), f"Wrong token amount (requested {TEST_GEN_TOKEN_COUNT})"
 
     y = model.raw_generate(TEST_PROMPT, max_new=TEST_GEN_TOKEN_COUNT, seed=TEST_SEED)
 
@@ -48,6 +51,7 @@ def test_torch_inference() -> None:
 
     print(x)
 
+
 # Horde
 def test_horde_load() -> None:
     global model
@@ -56,17 +60,22 @@ def test_horde_load() -> None:
     model = HordeInferenceModel()
     model.load()
 
+
 def test_horde_inference() -> None:
     x = model.raw_generate(TEST_PROMPT, max_new=TEST_GEN_TOKEN_COUNT, seed=TEST_SEED)
-    assert len(x.encoded[0]) == TEST_GEN_TOKEN_COUNT, f"Wrong token amount (requested {TEST_GEN_TOKEN_COUNT})"
+    assert (
+        len(x.encoded[0]) == TEST_GEN_TOKEN_COUNT
+    ), f"Wrong token amount (requested {TEST_GEN_TOKEN_COUNT})"
     print(x)
 
+
 # API
 def test_api_load() -> None:
     global model
     model = APIInferenceModel()
     model.load()
 
+
 def test_api_inference() -> None:
     x = model.raw_generate(TEST_PROMPT, max_new=TEST_GEN_TOKEN_COUNT, seed=TEST_SEED)
     # NOTE: Below test is flakey due to Horde worker-defined constraints

modeling/tokenizer.py

@@ -28,4 +28,3 @@ class GenericTokenizer:
             tokens = [tokens]
 
         return self.tokenizer.decode(tokens)
-