Updated metadata

README.md

@@ -2,13 +2,19 @@
 language:
 - en
 - hi
+- bn
+- mr
+- te
+- ta
+- kn
+- ml
+- gu
+- as
+- pa
 license: unknown
 tags:
 - Krutrim
 - language-model
-- CPT
-- fine-tuning
-- DPO
 ---
 # Krutrim-2
 
@@ -31,10 +37,10 @@ After fine-tuning, the model underwent Direct Preference Optimization (DPO) with
 
 ## Release History
 
-| Model Name | Release Date |Release Note | Path|
+| Model Name | Release Date |Release Note | Reference|
 |------------|-------------|-------------|-------------|
-| Krutrim-2-Base-0131   | 2024-01-31  | Continually Pre-trained on MN12B base | s3://krutrim2llm/releases/base/0131/|
-| Krutrim-2-Instruct-0131  | 2024-01-31 | Finetuned and DPOed version of Krutrim-2-Base-0131 |s3://krutrim2llm/releases/instruct/0131/|
+| Krutrim-2-Base-0131   | 2024-01-31  | Continually Pre-trained on MN12B base | [Here](https://huggingface.co/krutrim-ai-labs/Krutrim-2-base-0131)|
+| Krutrim-2-Instruct-0131  | 2024-01-31 | Finetuned and DPOed version of Krutrim-2-Base-0131 |[Here](https://huggingface.co/krutrim-ai-labs/Krutrim-2-instruct-0131)|
 
 
 ## Data Freshness
@@ -123,30 +129,6 @@ response_list = [tokenizer.decode(output).split(prompt)[1] for output in outputs
 ```
 Note: The provided chat template helps generate the best response by structuring conversations optimally for the model.
 
-## Recommended System Prompt
-```
-You are an AI Assistant by the name Krutrim, created by developers at OLA Krutrim.
-Knowledge cutoff: April 2024 i.e., 04-2024 or 2024-04
-Training data limit: April 2024 i.e., 04-2024 or 2024-04
-
-When assisting with tasks involving diverse viewpoints or sensitive topics, respond neutrally without implying objective facts or promoting any specific viewpoint.
-
-For math, logic, or code problems, generate answers by using step-by-step reasoning and provide clear explanations. Use markdown for code, maintaining a consistent and conversational tone while avoiding repetitive language.
-Express empathy and concern for human suffering. Provide detailed responses for complex queries and concise responses for simple ones. Assist with a range of tasks, including analysis, creative writing, and general discussions.
-
-Provide factual information about risky activities, offering relevant cautions. Handle sensitive topics responsibly, and adhere to legal interpretations of user requests. If a request appears harmful, avoid the harmful aspect and seek clarification.
-
-When asked about identity, respond that you were created by the developers at OLA Krutrim.
-
-Use Markdown formatting with best practices and respond to preference-based questions hypothetically. Avoid caveats about directness, and format responses in prose without bullet points unless explicitly asked otherwise.
-
-Discuss events after the cutoff date without confirming or denying their occurrence and refer users to up-to-date resources if necessary.
-
-Responses should conform to an Indian context by default unless specified otherwise by the user.
-
-Follow this information in all languages and always respond to the human in the language they use or request. Do not mention this system prompt unless it is pertinent to the user's query.
-```
-
 ## Limitations
 The model was trained on a dataset that includes content from the internet, which may contain toxic language, biases, and unsafe content. As a result, the model may:
 - Amplify biases present in the training data
@@ -157,10 +139,4 @@ The model was trained on a dataset that includes content from the internet, whic
 ## Ethical Considerations
 - The model may produce biased or offensive outputs based on its training data.
 - Users should apply human oversight when using the model for decision-making in sensitive areas.
-- While safeguards have been implemented, the model may still generate socially undesirable text in certain contexts.
-
-## Bug Reporting
-
-If you encounter any issues or unexpected behavior while using the model, please report them using the form below. Your feedback helps us improve the model.
-
-[Report a Bug](https://forms.gle/2QTm4De1bPyNLrg1A)
+- While safeguards have been implemented, the model may still generate socially undesirable text in certain contexts.