Spaces:

visualisable-ai
/

api

Sleeping

App Files Files Community

api / backend

384 kB

3 contributors

History: 55 commits

gary-boon

Limit QKV matrices to top 5 heads per layer to reduce response size

decb5ab 2 months ago

Dockerfile
460 Bytes

Fix backend structure - remove duplicates 6 months ago
architectural_analysis.py
10.6 kB

Add research attention analysis endpoints with Q/K/V extraction 3 months ago
attention_analysis.py
16.8 kB

Add research attention analysis endpoints with Q/K/V extraction 3 months ago
auth.py
1.18 kB

Fix backend structure - remove duplicates 6 months ago
context_efficiency_analyzer.py
13.4 kB

Add backend support for ICL emergence analysis 6 months ago
icl_attention_extractor.py
11.4 kB

Add Code Llama 7B support with hardware-aware filtering and ICL timeout fixes 4 months ago
icl_service.py
13.2 kB

Add Code Llama 7B support with hardware-aware filtering and ICL timeout fixes 4 months ago
induction_head_detector.py
13.8 kB

Add Code Llama 7B support with hardware-aware filtering and ICL timeout fixes 4 months ago
instrumentation.py
15.9 kB

Add research attention analysis endpoints with Q/K/V extraction 3 months ago
mistral_tokenizer.py
4.25 kB

Integrate mistral-common for correct Devstral tokenization 2 months ago
model_adapter.py
11 kB

Phase 2: Add Devstral backend support 2 months ago
model_config.py
5.95 kB

feat: Implement tier-based model filtering by device type 2 months ago
model_service.py
156 kB

Limit QKV matrices to top 5 heads per layer to reduce response size 2 months ago
pipeline_analyzer.py
29 kB

Add Code Llama 7B support with hardware-aware filtering and ICL timeout fixes 4 months ago
prompt_formatter.py
4.59 kB

Add tokenSections boundaries and update system prompt 2 months ago
qkv_extractor.py
20.7 kB

Add Code Llama 7B support with hardware-aware filtering and ICL timeout fixes 4 months ago
requirements.txt
166 Bytes

Fix backend structure - remove duplicates 6 months ago
storage.py
12.4 kB

Make zarr/numcodecs imports optional for ARM64 compatibility 2 months ago
swe_bench_service.py
18.6 kB

Fix pyarrow compatibility issue with datasets library 5 months ago
test_ablation.py
15.6 kB

Add ablation support to model service with comprehensive testing 6 months ago
tokenizer_utils.py
8.61 kB

Add research attention analysis endpoints with Q/K/V extraction 3 months ago