Unverified50% confidenceFactExact time
After 50 practice runs, Qwen's capped answers dropped from 93% to 33%, the math score rose from 7.9 to 48.1, elementary word problems rose from 14 to 67, and the instruction-following test rose from 21 to 62; when given only 10 or 25 practice runs, the loop barely faded and the math score remained low.
1
Sources
50%
Confidence
Long-term
Relevance
11/4/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn the experiment, Astra modified the weights of the small Qwen model, raising the MATH500 score from 7.9% to 48.1%, while the share of answers hitting the trial-run limit fell from 93% to 33%.71% similarUnverifiedQwen2.5 0.5B, across 96 replies, succeeded in direct parsing 0 times, succeeded via the parsing ladder 4 times, and conformed to the schema 0 times; most greedy replies echoed the user message and then entered a loop.71% similarUnverifiedOn 85 trained open-source-weight Qwen and Llama adapters, LoRAForge-S's lower-bound quality configuration retained only about 7.7% of the adapter parameters, while the overall average score dropped by at most 1.2 points on a 0–100 scale.70% similarUnverifiedOn math tasks, as the training answers become increasingly absurd, the proportion of misaligned responses soars from 4% to over 60%, while the model's general capabilities show almost no loss.68% similarUnverifiedQwen 3.8 is slower than 3.6, with prefill and inference speeds dropping by approximately 40% and 20% respectively.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1135747API
curl https://kongchang.com/api/v1/knowledge/claims/1135747MCP
get_claim(id=1135747)