feat: add jamba chat_template (#1843)

* feat: add jamba chat_template

* fix: black

* feat: jamba fsdp+qlora

---------

Co-authored-by: Gal Cohen <galc@ai21.com>

examples/jamba/qlora_fsdp.yaml

@@ -0,0 +1,61 @@
+base_model: ai21labs/Jamba-v0.1
+tokenizer_type: AutoTokenizer
+
+load_in_4bit: true
+strict: false
+use_tensorboard: true
+datasets:
+  - path: cgato/SlimOrcaDedupCleaned
+    type: chat_template
+    chat_template: jamba
+    drop_system_message: true
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.0
+output_dir: jamba-fsdp-qlora-ft
+save_safetensors: true
+adapter: qlora
+sequence_len: 2048
+sample_packing: true
+pad_to_sequence_len: true
+
+lora_r: 16
+lora_alpha: 16
+lora_dropout: 0.05
+lora_target_modules: [down_proj,gate_proj,in_proj,k_proj,o_proj,out_proj,q_proj,up_proj,v_proj,x_proj]
+lora_target_linear: false
+
+gradient_accumulation_steps: 4
+micro_batch_size: 1
+num_epochs: 2
+optimizer: adamw_torch
+lr_scheduler: cosine
+learning_rate: 0.00001
+
+train_on_inputs: false
+group_by_length: false
+bf16: true
+tf32: true
+
+gradient_checkpointing: true
+gradient_checkpointing_kwargs:
+  use_reentrant: true
+logging_steps: 1
+flash_attention: true
+
+warmup_steps: 10
+evals_per_epoch: 1
+saves_per_epoch: 1
+weight_decay: 0.0
+fsdp:
+  - full_shard
+  - auto_wrap
+fsdp_config:
+  fsdp_limit_all_gathers: true
+  fsdp_sync_module_states: true
+  fsdp_offload_params: false
+  fsdp_use_orig_params: false
+  fsdp_cpu_ram_efficient_loading: true
+  fsdp_auto_wrap_policy: TRANSFORMER_BASED_WRAP
+  fsdp_transformer_layer_cls_to_wrap: JambaAttentionDecoderLayer,JambaMambaDecoderLayer
+  fsdp_state_dict_type: FULL_STATE_DICT
+  fsdp_sharding_strategy: FULL_SHARD

src/axolotl/utils/config/models/input/v0_4_1/__init__.py

@@ -190,6 +190,7 @@ class ChatTemplate(str, Enum):
     llama3 = "llama3"  # pylint: disable=invalid-name
     phi_3 = "phi_3"  # pylint: disable=invalid-name
     deepseek_v2 = "deepseek_v2"  # pylint: disable=invalid-name
+    jamba = "jamba"  # pylint: disable=invalid-name
 
 
 class LoftQConfig(BaseModel):