fixes w/ example for super basic lora starter

examples/lora-alpaca-7b/config.yml

@@ -0,0 +1,67 @@
+base_model: huggyllama/llama-7b
+base_model_config: huggyllama/llama-7b
+model_type: LlamaForCausalLM
+tokenizer_type: LlamaTokenizer
+load_in_8bit: true
+load_in_4bit: false
+strict: false
+push_dataset_to_hub:
+datasets:
+  - path: teknium/GPT4-LLM-Cleaned
+    type: alpaca
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.02
+adapter: lora
+lora_model_dir:
+sequence_len: 512
+max_packed_sequence_len:
+lora_r: 8
+lora_alpha: 16
+lora_dropout: 0.0
+lora_target_modules:
+  - gate_proj
+  - down_proj
+  - up_proj
+  - q_proj
+  - v_proj
+  - k_proj
+  - o_proj
+lora_fan_in_fan_out:
+wandb_project:
+wandb_watch:
+wandb_run_id:
+wandb_log_model:
+output_dir: ./lora-out
+batch_size: 4
+micro_batch_size: 1
+num_epochs: 4
+optimizer: adamw_bnb_8bit
+torchdistx_path:
+lr_scheduler: cosine
+learning_rate: 0.0002
+train_on_inputs: false
+group_by_length: false
+bf16: false
+fp16: true
+tf32: true
+gradient_checkpointing: true
+early_stopping_patience:
+resume_from_checkpoint:
+local_rank:
+logging_steps: 1
+xformers_attention: true
+flash_attention:
+gptq_groupsize:
+gptq_model_v1:
+warmup_steps: 10
+eval_steps: 50
+save_steps:
+debug:
+deepspeed:
+weight_decay: 0.0
+fsdp:
+fsdp_config:
+special_tokens:
+  bos_token: "<s>"
+  eos_token: "</s>"
+  unk_token: "<unk>"

src/axolotl/prompters.py

@@ -18,7 +18,7 @@ class AlpacaPrompter:
     prompt_style = None
 
     def __init__(self, prompt_style="instruct"):
-        self.prompt_style = prompt_style
+        self.prompt_style = prompt_style if prompt_style else PromptStyle.instruct.value
         self.match_prompt_style()
 
     def match_prompt_style(self):

src/axolotl/utils/data.py

@@ -60,10 +60,12 @@ def load_tokenized_prepared_datasets(
         else Path(default_dataset_prepared_path) / ds_hash
     )
     dataset = None
+    use_auth_token = False
     try:
         if cfg.push_dataset_to_hub:
+            use_auth_token = True
             dataset = load_dataset(
-                f"{cfg.push_dataset_to_hub}/{ds_hash}", use_auth_token=True
+                f"{cfg.push_dataset_to_hub}/{ds_hash}", use_auth_token=use_auth_token
             )
             dataset = dataset["train"]
     except:
@@ -83,7 +85,7 @@ def load_tokenized_prepared_datasets(
             ds = None
             ds_from_hub = False
             try:
-                load_dataset(d.path, streaming=True, use_auth_token=True)
+                load_dataset(d.path, streaming=True, use_auth_token=use_auth_token)
                 ds_from_hub = True
             except FileNotFoundError:
                 pass
@@ -99,10 +101,10 @@ def load_tokenized_prepared_datasets(
                         d.path,
                         streaming=False,
                         data_files=d.data_files,
-                        use_auth_token=True,
+                        use_auth_token=use_auth_token,
                     )
                 else:
-                    ds = load_dataset(d.path, streaming=False, use_auth_token=True)
+                    ds = load_dataset(d.path, streaming=False, use_auth_token=use_auth_token)
             else:
                 fp = hf_hub_download(
                     repo_id=d.path, repo_type="dataset", filename=d.data_files