use fastchat conversations template (#578)

* use fastchat conversations template

* require fastchat (fschat) pip install

* handle roles dynamically from conversation

* tweak fastchat conversation with a monkeypatch to get individual turns

* fix up so it works with multiple conversation styles, and don't strip the turns

* fix sharegpt fixture now that we're using a more correct tokenization

* use a new prompter and support fastchat conversation type

* use sharegpt from prompt strategies now

* update docs, add chatml template

* add a newline after im_end token

* ensure we correctly set system message

* update per PR feedback to handle deprecated sharegpt types

* don't add duplicate wandb req

* make sharegpt fields configurable from yml

* llama2 fixes

* don't fail fatally when turns are improper

tests/test_validation.py

@@ -374,3 +374,26 @@ class ValidationTest(unittest.TestCase):
         )
 
         validate_config(cfg)
+
+    def test_sharegpt_deprecation(self):
+        cfg = DictDefault(
+            {"datasets": [{"path": "lorem/ipsum", "type": "sharegpt:chat"}]}
+        )
+        with self._caplog.at_level(logging.WARNING):
+            validate_config(cfg)
+            assert any(
+                "`type: sharegpt:chat` will soon be deprecated." in record.message
+                for record in self._caplog.records
+            )
+        assert cfg.datasets[0].type == "sharegpt"
+
+        cfg = DictDefault(
+            {"datasets": [{"path": "lorem/ipsum", "type": "sharegpt_simple:load_role"}]}
+        )
+        with self._caplog.at_level(logging.WARNING):
+            validate_config(cfg)
+            assert any(
+                "`type: sharegpt_simple` will soon be deprecated." in record.message
+                for record in self._caplog.records
+            )
+        assert cfg.datasets[0].type == "sharegpt:load_role"