monsterapi
/

llama2_7b_norobots

Model card Files Files and versions Community

souvik0306 commited on Nov 20, 2023

Commit

e843c58

•

1 Parent(s): c4a1d81

Update README.md

Files changed (1) hide show

README.md +5 -10

README.md CHANGED Viewed

@@ -35,22 +35,17 @@ With the utilization of [MonsterAPI](https://monsterapi.ai)'s [LLM finetuner](ht
 - **Total Finetuning Cost:** $1.313
 - **Model Path:** meta-llama/Llama-2-7b-hf
 - **Learning Rate:** 0.0002
-- **Data Split:** 99% train 1% validation
 - **Gradient Accumulation Steps:** 4
 - **lora r:** 32
 - **lora alpha:** 64
----
-Prompt Structure
 ```
-### INSTRUCTION:
-[instruction]
-### RESPONSE:
-[text]
 ```
-Train loss :
-![eval loss](https://cdn-uploads.huggingface.co/production/uploads/63ba46aa0a9866b28cb19a14/D84FFl8hAorzJbtSfiiIT.png)
 license: apache-2.0

 - **Total Finetuning Cost:** $1.313
 - **Model Path:** meta-llama/Llama-2-7b-hf
 - **Learning Rate:** 0.0002
+- **Data Split:** 100% train
 - **Gradient Accumulation Steps:** 4
 - **lora r:** 32
 - **lora alpha:** 64
+#### Prompt Structure
 ```
+<|system|> <|endoftext|> <|user|> [USER PROMPT]<|endoftext|> <|assistant|> [ASSISTANT ANSWER] <|endoftext|>
 ```
+#### Train loss :
+![eval loss](https://cdn-uploads.huggingface.co/production/uploads/63ba46aa0a9866b28cb19a14/_UwicIoHhj1RrMjt_63vQ.png)
 license: apache-2.0