x1 | BRAHMAI
7 Billion Parameters | v0.1
Introducing x1-7B-v0.1, the first Large Language Model by BRAHMAI with built-in reasoning capabilities. Based on the powerful Qwen2.5 7B architecture, this model has been fine-tuned using a synthetic dataset generated from open-source models across multiple datasets and custom agentic pipelines.
Note:
As the first iteration of this model, x1-7B-v0.1 demonstrates strong performance in answering reasoning-based questions, making it a valuable tool for advancing further development. When reasoning, the model encapsulates its thought process within <thinking> and </thinking> tags before delivering the final response.
Known Limitations
While promising, this model has its quirks:
- Unexpected Behavior with Simple Inputs: For instance, it may break when simply greeted with "Hi."
- Excessive Reasoning: Even for basic queries, the model often engages in elaborate reasoning chains.
- Repetitive Responses: Occasionally, after providing an answer, the model gets stuck in a loop, repeating patterns.
Example Conversations
Note: All examples were tested with max_new_tokens set to 512.
Example 1: Comparing 9.24 and 9.8
USER: What is larger? 9.24 or 9.8?
ASSISTANT: <thinking>
To determine which number is larger between 9.24 and 9.8, we can compare the digits in each place value.
1. Both numbers have the same digit in the ones place (9).
2. In the tenths place, 9.24 has a 2, while 9.8 has an 8. Since 8 is greater than 2, 9.8 is larger than 9.24.
Thus, 9.8 is larger than 9.24.
</thinking>
9.8 is larger than 9.24.
Example 2: The "Hi" Scenario
USER: Hi
ASSISTANT: <thinking>
The user is asking for a list of 100 words that start with the letter "S" and end with the letter "e". To generate this list, I need to think of words that fit these criteria.
I will start by brainstorming words that begin with "S" and end with "e". Some examples include "sage", "sneeze", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve", "sieve",
Example 3: Counting "R"s in "STRAWBERRY"
USER: How many R's are in STRAWBERRY?
ASSISTANT: <thinking>
To determine the number of "R"s in "STRAWBERRY," I’ll count the occurrences of "R."
The word "STRAWBERRY" consists of the following letters: S, T, R, A, W, B, E, R, R, Y.
Counting the "R"s:
- The first "R" is in the third position.
- The second "R" is in the eighth position.
- The third "R" is in the ninth position.
Therefore, there are 3 "R"s in "STRAWBERRY."
</thinking>
There are 3 "R"s in the word "STRAWBERRY."
Moving Forward
We're committed to refining this model. Stay tuned for improvements as we continue to fine-tune it.
If you have any questions or feedback, feel free to reach out to us at hello@brahmai.in.