RishuD7/Llama3.1_question_answer_model_1

README.md

@@ -18,7 +18,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [meta-llama/Meta-Llama-3.1-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.1-8B-Instruct) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.2091
+- Loss: 0.2020
 
 ## Model description
 
@@ -43,9 +43,9 @@ The following `bitsandbytes` quantization config was used during training:
 - llm_int8_skip_modules: None
 - llm_int8_enable_fp32_cpu_offload: False
 - llm_int8_has_fp16_weight: False
-- bnb_4bit_quant_type: fp4
+- bnb_4bit_quant_type: nf4
 - bnb_4bit_use_double_quant: False
-- bnb_4bit_compute_dtype: float32
+- bnb_4bit_compute_dtype: bfloat16
 - bnb_4bit_quant_storage: uint8
 - load_in_4bit: False
 - load_in_8bit: True
@@ -66,9 +66,9 @@ The following hyperparameters were used during training:
 
 | Training Loss | Epoch  | Step | Validation Loss |
 |:-------------:|:------:|:----:|:---------------:|
-| 0.1235        | 0.9991 | 583  | 0.1561          |
-| 0.0837        | 2.0    | 1167 | 0.1912          |
-| 0.0693        | 2.9974 | 1749 | 0.2091          |
+| 0.0968        | 0.9996 | 1150 | 0.1602          |
+| 0.076         | 2.0    | 2301 | 0.1828          |
+| 0.0678        | 2.9987 | 3450 | 0.2020          |
 
 
 ### Framework versions