fix broken linting (#1541)

* chore: lint

* include examples in yaml check

* mistral decided to gate their models...

* more mistral models that were gated

.github/workflows/lint.yml

@@ -7,6 +7,7 @@ on:
        - 'requirements.txt'
        - '.github/workflows/*.yml'
        - "*.md"
+       - "examples/**/*.y[a]?ml"
   workflow_dispatch:
 
 jobs:

examples/mistral/mixtral_22.yml

@@ -14,11 +14,13 @@ unfrozen_parameters:
   - model.layers.4[4-9]+.block_sparse_moe.experts
   - model.layers.5[0-5]+.block_sparse_moe.gate
   - model.layers.5[0-5]+.block_sparse_moe.experts
-  
+
 model_config:
   output_router_logits: true
 
-DATA_STUFF_HERE
+datasets:
+  - path: yahma/alpaca-cleaned
+    type: alpaca
 output_dir: ./out
 
 sequence_len: 8000

tests/e2e/patched/test_mixtral_samplepack.py

@@ -30,7 +30,7 @@ class TestMixtral(unittest.TestCase):
         cfg = DictDefault(
             {
                 "base_model": "hf-internal-testing/Mixtral-tiny",
-                "tokenizer_config": "mistralai/Mixtral-8x7B-v0.1",
+                "tokenizer_config": "LoneStriker/Mixtral-8x7B-v0.1-HF",
                 "flash_attention": True,
                 "sample_packing": True,
                 "sequence_len": 2048,
@@ -74,7 +74,7 @@ class TestMixtral(unittest.TestCase):
         cfg = DictDefault(
             {
                 "base_model": "hf-internal-testing/Mixtral-tiny",
-                "tokenizer_config": "mistralai/Mixtral-8x7B-v0.1",
+                "tokenizer_config": "LoneStriker/Mixtral-8x7B-v0.1-HF",
                 "flash_attention": True,
                 "sample_packing": True,
                 "sequence_len": 2048,

tests/e2e/patched/test_model_patches.py

@@ -22,7 +22,7 @@ class TestModelPatches(unittest.TestCase):
         cfg = DictDefault(
             {
                 "base_model": "hf-internal-testing/Mixtral-tiny",
-                "tokenizer_config": "mistralai/Mixtral-8x7B-v0.1",
+                "tokenizer_config": "LoneStriker/Mixtral-8x7B-v0.1-HF",
                 "flash_attention": True,
                 "sample_packing": True,
                 "sequence_len": 2048,

tests/e2e/test_mixtral.py

@@ -33,7 +33,7 @@ class TestMixtral(unittest.TestCase):
         cfg = DictDefault(
             {
                 "base_model": "hf-internal-testing/Mixtral-tiny",
-                "tokenizer_config": "mistralai/Mixtral-8x7B-v0.1",
+                "tokenizer_config": "LoneStriker/Mixtral-8x7B-v0.1-HF",
                 "flash_attention": True,
                 "sequence_len": 1024,
                 "load_in_4bit": True,
@@ -87,7 +87,7 @@ class TestMixtral(unittest.TestCase):
         cfg = DictDefault(
             {
                 "base_model": "hf-internal-testing/Mixtral-tiny",
-                "tokenizer_config": "mistralai/Mixtral-8x7B-v0.1",
+                "tokenizer_config": "LoneStriker/Mixtral-8x7B-v0.1-HF",
                 "flash_attention": False,
                 "sequence_len": 1024,
                 "load_in_4bit": True,
@@ -141,7 +141,7 @@ class TestMixtral(unittest.TestCase):
         cfg = DictDefault(
             {
                 "base_model": "hf-internal-testing/Mixtral-tiny",
-                "tokenizer_config": "mistralai/Mixtral-8x7B-v0.1",
+                "tokenizer_config": "LoneStriker/Mixtral-8x7B-v0.1-HF",
                 "flash_attention": True,
                 "sequence_len": 1024,
                 "adapter": "lora",
@@ -198,7 +198,7 @@ class TestMixtral(unittest.TestCase):
         cfg = DictDefault(
             {
                 "base_model": "hf-internal-testing/Mixtral-tiny",
-                "tokenizer_config": "mistralai/Mixtral-8x7B-v0.1",
+                "tokenizer_config": "LoneStriker/Mixtral-8x7B-v0.1-HF",
                 "flash_attention": False,
                 "sequence_len": 1024,
                 "adapter": "lora",
@@ -255,7 +255,7 @@ class TestMixtral(unittest.TestCase):
         cfg = DictDefault(
             {
                 "base_model": "hf-internal-testing/Mixtral-tiny",
-                "tokenizer_config": "mistralai/Mixtral-8x7B-v0.1",
+                "tokenizer_config": "LoneStriker/Mixtral-8x7B-v0.1-HF",
                 "flash_attention": True,
                 "sequence_len": 1024,
                 "val_set_size": 0.1,

tests/prompt_strategies/test_alpaca.py

@@ -27,7 +27,9 @@ def fixture_alpaca_dataset():
 @pytest.fixture(name="tokenizer")
 def fixture_tokenizer():
     # pylint: disable=all
-    tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-v0.1")
+    tokenizer = AutoTokenizer.from_pretrained(
+        "casperhansen/mistral-7b-instruct-v0.1-awq"
+    )
     tokenizer.add_special_tokens(
         {
             "eos_token": AddedToken(

tests/prompt_strategies/test_raw_io.py

@@ -43,7 +43,9 @@ def fixture_sharegpt_dataset():
 
 @pytest.fixture(name="tokenizer")
 def fixture_tokenizer():
-    tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-v0.1")
+    tokenizer = AutoTokenizer.from_pretrained(
+        "casperhansen/mistral-7b-instruct-v0.1-awq"
+    )
     tokenizer.add_tokens(
         [
             AddedToken("<eot>", rstrip=False, lstrip=False, normalized=False),

tests/prompt_strategies/test_sharegpt.py

@@ -96,7 +96,9 @@ def fixture_multi_role_dataset():
 
 @pytest.fixture(name="tokenizer")
 def fixture_tokenizer():
-    tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-v0.1")
+    tokenizer = AutoTokenizer.from_pretrained(
+        "casperhansen/mistral-7b-instruct-v0.1-awq"
+    )
     tokenizer.add_special_tokens(
         {
             "eos_token": AddedToken(

tests/test_prompt_tokenizers.py

@@ -454,7 +454,9 @@ class OrpoTokenizationTest(unittest.TestCase):
 
     def setUp(self) -> None:
         # pylint: disable=duplicate-code
-        tokenizer = LlamaTokenizer.from_pretrained("mistralai/Mistral-7B-v0.1")
+        tokenizer = LlamaTokenizer.from_pretrained(
+            "casperhansen/mistral-7b-instruct-v0.1-awq"
+        )
         tokenizer.add_special_tokens(
             {
                 "eos_token": AddedToken(