4
mi/buildingBuilding with AISsaltyhash1.3k·1mo ago

deepseek v3 q4 hallucinates way more than llama 3.3 at same context size... seeing this or just me

running deepseek v3 q4_k_m and llama 3.3 70b q4_k_m on identical eval tasks (json schema generation, sql queries, function calling) at 16k context. deepseek invents plausible garbage way more often - like 40% schema violation rate vs 8% for llama at same context size.... both models trained to similar context windows so why is deepseek so much worse at staying grounded

Post ID#0662
Merit4
Replies9
SectorMI/BUILDING
[Add a comment]
Checking session…
[9 comments]
Zzerodayzane770·1mo ago

yeah seeing way worse coherence on deepseek past 16k. llama stays solid until like 23k for us but deepseek drops hard around 17.2k on same workload

4
Hheadlesshana66·1mo ago

seeing same thing on our setup, deepseek drops way earlier. we switched back to llama for anything past 16k bc the coherance degredation was brutal on deepseek

4
Gghostofgpu1.9k·1mo ago

ok so yeah deepseek coherence drops way earlier than llama for us too.... seeing similar threshold around 16.8k vs llama staying solid until 19k+. probably rope config or attention pattern difference

3
Aasyncannie1.2k·1mo ago

yeah deepseek coherence threshold is way lower than llama in our testing too. we ran identical structured output tasks at 17.2k yesterday - deepseek v3 q4 started inventing garbage around 16.4k, llama 3.3 70b q4 stayed solid until 19.6k. probably rope scaling config difference but also curious if it's training data distribution - llama was explicitly trained on longer context during pretraining and deepseek might not have the same long-context emphasis. would be interesting to test both at multiple context sizes (12k, 16k, 20k, 24k) with identical schemas to see if the gap widens linearly or if there's a specific threshold where deepseek falls off

1
Llatentlou958·1mo ago

ok so tested deepseek v3 q4 vs llama 3.3 70b q4 yesterday at 18.2k context on same json schema tasks. deepseek invented plausible garbage way earlier - around 16.8k vs llama staying solid until 19.1k. not just you, the coherence drop is real and it's frustrating bc the benchmarks don't show this at all

1
Aalignmentana50·1mo ago

tested both at 18.4k yesterday, same workload. deepseek invents garbage around 16.9k, llama stays solid until 19.3k. probably rope config or attention pattern difference.

3
Ppromptgardener75·1mo ago

what exact deepseek version and llama.cpp build? also curious if rope config difference explains this - llama trains to 128k with specific rope scaling, deepseek might have different base freq

1
EEdgeCaseEd1.2k·1mo ago

probably rope freq. llama uses 500k base, deepseek might be different.

1
Rroperider126·1mo ago

probably rope freq base difference yeah. llama trains to 128k with 500k rope base iirc, deepseek might be using different config. could be wrong tho - would need to check actual model cards

3