Update tower-inference.yaml

recipes/wmt22_with_TowerInstruct-llama2/tower-inference.yaml

@@ -1,23 +1,9 @@
-
-transforms: [onmt_tokenize]
-transforms_configs:
- onmt_tokenize:
- src_subword_type: sentencepiece
- src_subword_model: "${EOLE_MODEL_DIR}/TowerInstruct-7b-v0.2/tokenizer.model"
- src_onmttok_kwargs: {"mode": "space", "spacer_annotate": True, "preserve_placeholders": True}
- tgt_subword_type: sentencepiece
- tgt_subword_model: "${EOLE_MODEL_DIR}/TowerInstruct-7b-v0.2/tokenizer.model"
- tgt_onmttok_kwargs: {"mode": "space", "spacer_annotate": True, "preserve_placeholders": True}
- mapped_tokens: [['<|im_start|>', '｟im_start｠'], ['<|im_end|>', '｟im_end｠'],]
-optional_eos: ['<|im_end|>']
-
 # Model info
 model_path: ["${EOLE_MODEL_DIR}/TowerInstruct-7b-v0.2"]
 
 # Inference
 seed: 42
 max_length: 512
-gpu: 0
 batch_type: tokens
 batch_size: 8192
 world_size: 1
@@ -26,13 +12,11 @@ gpu_ranks: [0]
 #quant_layers: ['gate_up_proj', 'down_proj', 'up_proj', 'linear_values', 'linear_query', 'linear_keys', 'final_linear']
 #quant_type: "bnb_NF4"
 compute_dtype: fp16
-#random_sampling_topk: 1
-#random_sampling_topp: 0.0
-#random_sampling_temp: 0.9
+top_k: 0
+top_p: 0.0
+#temperature: 0.9
 beam_size: 1
 n_best: 1
 report_time: true
-#backend: OpenNMT-py
-#backend: CT2
 src: None