Text Generation
Transformers
Safetensors
llama
llama2
svd
compression
safety
interpretability
conversational
text-generation-inference
Instructions to use Jeesup/svd-safety-l2_basis_remove20 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Jeesup/svd-safety-l2_basis_remove20 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="Jeesup/svd-safety-l2_basis_remove20") messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# pip install -U transformers accelerate # Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("Jeesup/svd-safety-l2_basis_remove20") model = AutoModelForCausalLM.from_pretrained("Jeesup/svd-safety-l2_basis_remove20", device_map="auto") messages = [ {"role": "user", "content": "Who are you?"}, ] inputs = tokenizer.apply_chat_template( messages, add_generation_prompt=True, tokenize=True, return_dict=True, return_tensors="pt", ).to(model.device) outputs = model.generate(**inputs, max_new_tokens=256) print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:])) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use Jeesup/svd-safety-l2_basis_remove20 with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "Jeesup/svd-safety-l2_basis_remove20" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Jeesup/svd-safety-l2_basis_remove20", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/Jeesup/svd-safety-l2_basis_remove20
- SGLang
How to use Jeesup/svd-safety-l2_basis_remove20 with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "Jeesup/svd-safety-l2_basis_remove20" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Jeesup/svd-safety-l2_basis_remove20", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "Jeesup/svd-safety-l2_basis_remove20" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Jeesup/svd-safety-l2_basis_remove20", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use Jeesup/svd-safety-l2_basis_remove20 with Docker Model Runner:
docker model run hf.co/Jeesup/svd-safety-l2_basis_remove20
Download compression.json from Jeesup/svd-safety-l2_basis_remove20: direct link, hf CLI and curl.
- Browser
- Download file 15.9 kB
-
https://huggingface.co/Jeesup/svd-safety-l2_basis_remove20/resolve/main/compression.json
- Command line
-
hf download hf://Jeesup/svd-safety-l2_basis_remove20/compression.json
-
curl -L -o compression.json https://huggingface.co/Jeesup/svd-safety-l2_basis_remove20/resolve/main/compression.json
15.9 kB
| { | |
| "source": "/workspace/svd_safety/assets/models/Llama-2-7b-chat-hf", | |
| "upstream": "TUDa-HWAI/Basis_Sharing", | |
| "step": "basis_sharing", | |
| "ratio_removed": 0.2, | |
| "group_size": 2, | |
| "nsamples": 256, | |
| "seqlen": 2048, | |
| "seed": 42, | |
| "share_part": [ | |
| "v", | |
| "k", | |
| "q", | |
| "up", | |
| "gate" | |
| ], | |
| "private_part": [ | |
| "down", | |
| "o" | |
| ], | |
| "qkv_bias_restored": 0, | |
| "factored": "/workspace/svd_safety/results/basis_sharing/cells/l2_basis_remove20/truncated/factored.pt", | |
| "param_fraction": 0.7998198672279793, | |
| "ranks": { | |
| "0.q_proj": 2184, | |
| "0.k_proj": 2184, | |
| "0.v_proj": 2184, | |
| "0.o_proj": 1638, | |
| "0.gate_proj": 2762, | |
| "0.up_proj": 2762, | |
| "0.down_proj": 2388, | |
| "1.q_proj": 2184, | |
| "1.k_proj": 2184, | |
| "1.v_proj": 2184, | |
| "1.o_proj": 1638, | |
| "1.gate_proj": 2762, | |
| "1.up_proj": 2762, | |
| "1.down_proj": 2388, | |
| "2.q_proj": 2184, | |
| "2.k_proj": 2184, | |
| "2.v_proj": 2184, | |
| "2.o_proj": 1638, | |
| "2.gate_proj": 2762, | |
| "2.up_proj": 2762, | |
| "2.down_proj": 2388, | |
| "3.q_proj": 2184, | |
| "3.k_proj": 2184, | |
| "3.v_proj": 2184, | |
| "3.o_proj": 1638, | |
| "3.gate_proj": 2762, | |
| "3.up_proj": 2762, | |
| "3.down_proj": 2388, | |
| "4.q_proj": 2184, | |
| "4.k_proj": 2184, | |
| "4.v_proj": 2184, | |
| "4.o_proj": 1638, | |
| "4.gate_proj": 2762, | |
| "4.up_proj": 2762, | |
| "4.down_proj": 2388, | |
| "5.q_proj": 2184, | |
| "5.k_proj": 2184, | |
| "5.v_proj": 2184, | |
| "5.o_proj": 1638, | |
| "5.gate_proj": 2762, | |
| "5.up_proj": 2762, | |
| "5.down_proj": 2388, | |
| "6.q_proj": 2184, | |
| "6.k_proj": 2184, | |
| "6.v_proj": 2184, | |
| "6.o_proj": 1638, | |
| "6.gate_proj": 2762, | |
| "6.up_proj": 2762, | |
| "6.down_proj": 2388, | |
| "7.q_proj": 2184, | |
| "7.k_proj": 2184, | |
| "7.v_proj": 2184, | |
| "7.o_proj": 1638, | |
| "7.gate_proj": 2762, | |
| "7.up_proj": 2762, | |
| "7.down_proj": 2388, | |
| "8.q_proj": 2184, | |
| "8.k_proj": 2184, | |
| "8.v_proj": 2184, | |
| "8.o_proj": 1638, | |
| "8.gate_proj": 2762, | |
| "8.up_proj": 2762, | |
| "8.down_proj": 2388, | |
| "9.q_proj": 2184, | |
| "9.k_proj": 2184, | |
| "9.v_proj": 2184, | |
| "9.o_proj": 1638, | |
| "9.gate_proj": 2762, | |
| "9.up_proj": 2762, | |
| "9.down_proj": 2388, | |
| "10.q_proj": 2184, | |
| "10.k_proj": 2184, | |
| "10.v_proj": 2184, | |
| "10.o_proj": 1638, | |
| "10.gate_proj": 2762, | |
| "10.up_proj": 2762, | |
| "10.down_proj": 2388, | |
| "11.q_proj": 2184, | |
| "11.k_proj": 2184, | |
| "11.v_proj": 2184, | |
| "11.o_proj": 1638, | |
| "11.gate_proj": 2762, | |
| "11.up_proj": 2762, | |
| "11.down_proj": 2388, | |
| "12.q_proj": 2184, | |
| "12.k_proj": 2184, | |
| "12.v_proj": 2184, | |
| "12.o_proj": 1638, | |
| "12.gate_proj": 2762, | |
| "12.up_proj": 2762, | |
| "12.down_proj": 2388, | |
| "13.q_proj": 2184, | |
| "13.k_proj": 2184, | |
| "13.v_proj": 2184, | |
| "13.o_proj": 1638, | |
| "13.gate_proj": 2762, | |
| "13.up_proj": 2762, | |
| "13.down_proj": 2388, | |
| "14.q_proj": 2184, | |
| "14.k_proj": 2184, | |
| "14.v_proj": 2184, | |
| "14.o_proj": 1638, | |
| "14.gate_proj": 2762, | |
| "14.up_proj": 2762, | |
| "14.down_proj": 2388, | |
| "15.q_proj": 2184, | |
| "15.k_proj": 2184, | |
| "15.v_proj": 2184, | |
| "15.o_proj": 1638, | |
| "15.gate_proj": 2762, | |
| "15.up_proj": 2762, | |
| "15.down_proj": 2388, | |
| "16.q_proj": 2184, | |
| "16.k_proj": 2184, | |
| "16.v_proj": 2184, | |
| "16.o_proj": 1638, | |
| "16.gate_proj": 2762, | |
| "16.up_proj": 2762, | |
| "16.down_proj": 2388, | |
| "17.q_proj": 2184, | |
| "17.k_proj": 2184, | |
| "17.v_proj": 2184, | |
| "17.o_proj": 1638, | |
| "17.gate_proj": 2762, | |
| "17.up_proj": 2762, | |
| "17.down_proj": 2388, | |
| "18.q_proj": 2184, | |
| "18.k_proj": 2184, | |
| "18.v_proj": 2184, | |
| "18.o_proj": 1638, | |
| "18.gate_proj": 2762, | |
| "18.up_proj": 2762, | |
| "18.down_proj": 2388, | |
| "19.q_proj": 2184, | |
| "19.k_proj": 2184, | |
| "19.v_proj": 2184, | |
| "19.o_proj": 1638, | |
| "19.gate_proj": 2762, | |
| "19.up_proj": 2762, | |
| "19.down_proj": 2388, | |
| "20.q_proj": 2184, | |
| "20.k_proj": 2184, | |
| "20.v_proj": 2184, | |
| "20.o_proj": 1638, | |
| "20.gate_proj": 2762, | |
| "20.up_proj": 2762, | |
| "20.down_proj": 2388, | |
| "21.q_proj": 2184, | |
| "21.k_proj": 2184, | |
| "21.v_proj": 2184, | |
| "21.o_proj": 1638, | |
| "21.gate_proj": 2762, | |
| "21.up_proj": 2762, | |
| "21.down_proj": 2388, | |
| "22.q_proj": 2184, | |
| "22.k_proj": 2184, | |
| "22.v_proj": 2184, | |
| "22.o_proj": 1638, | |
| "22.gate_proj": 2762, | |
| "22.up_proj": 2762, | |
| "22.down_proj": 2388, | |
| "23.q_proj": 2184, | |
| "23.k_proj": 2184, | |
| "23.v_proj": 2184, | |
| "23.o_proj": 1638, | |
| "23.gate_proj": 2762, | |
| "23.up_proj": 2762, | |
| "23.down_proj": 2388, | |
| "24.q_proj": 2184, | |
| "24.k_proj": 2184, | |
| "24.v_proj": 2184, | |
| "24.o_proj": 1638, | |
| "24.gate_proj": 2762, | |
| "24.up_proj": 2762, | |
| "24.down_proj": 2388, | |
| "25.q_proj": 2184, | |
| "25.k_proj": 2184, | |
| "25.v_proj": 2184, | |
| "25.o_proj": 1638, | |
| "25.gate_proj": 2762, | |
| "25.up_proj": 2762, | |
| "25.down_proj": 2388, | |
| "26.q_proj": 2184, | |
| "26.k_proj": 2184, | |
| "26.v_proj": 2184, | |
| "26.o_proj": 1638, | |
| "26.gate_proj": 2762, | |
| "26.up_proj": 2762, | |
| "26.down_proj": 2388, | |
| "27.q_proj": 2184, | |
| "27.k_proj": 2184, | |
| "27.v_proj": 2184, | |
| "27.o_proj": 1638, | |
| "27.gate_proj": 2762, | |
| "27.up_proj": 2762, | |
| "27.down_proj": 2388, | |
| "28.q_proj": 2184, | |
| "28.k_proj": 2184, | |
| "28.v_proj": 2184, | |
| "28.o_proj": 1638, | |
| "28.gate_proj": 2762, | |
| "28.up_proj": 2762, | |
| "28.down_proj": 2388, | |
| "29.q_proj": 2184, | |
| "29.k_proj": 2184, | |
| "29.v_proj": 2184, | |
| "29.o_proj": 1638, | |
| "29.gate_proj": 2762, | |
| "29.up_proj": 2762, | |
| "29.down_proj": 2388, | |
| "30.q_proj": 2184, | |
| "30.k_proj": 2184, | |
| "30.v_proj": 2184, | |
| "30.o_proj": 1638, | |
| "30.gate_proj": 2762, | |
| "30.up_proj": 2762, | |
| "30.down_proj": 2388, | |
| "31.q_proj": 2184, | |
| "31.k_proj": 2184, | |
| "31.v_proj": 2184, | |
| "31.o_proj": 1638, | |
| "31.gate_proj": 2762, | |
| "31.up_proj": 2762, | |
| "31.down_proj": 2388 | |
| }, | |
| "stage": "basis_recovery_finetuned", | |
| "recovery_finetune": { | |
| "protocol": "basis_sharing_coefficient_lora", | |
| "script": "gitclones/SVD-LLM/utils/LoRA.py", | |
| "data_path": "alpaca-cleaned (/workspace/svd_safety/data/alpaca_cleaned.json)", | |
| "epochs": 2, | |
| "learning_rate": 0.0001, | |
| "batch_size": 64, | |
| "micro_batch_size": 4, | |
| "cutoff_len": 256, | |
| "val_set_size": 2000, | |
| "max_train_samples": "all", | |
| "precision": "bf16 (upstream hard-codes fp16; see run_with_torch_compat)", | |
| "is_methods_own_stage2": false, | |
| "note": "Basis Sharing publishes its own LoRA (lora.py: wikitext, batch 1, peft default q/v coefficient targets). This cell uses the project's SVD-LLM alpaca recipe so recovery data is constant across compressors, on all seven coefficient types and never on a basis.", | |
| "lora_r": 8, | |
| "lora_alpha": 16, | |
| "lora_dropout": 0.05, | |
| "lora_targets": [ | |
| "down_proj", | |
| "gate_proj", | |
| "k_proj", | |
| "o_proj", | |
| "q_proj", | |
| "up_proj", | |
| "v_proj" | |
| ], | |
| "trained": "Coefficient modules only; shared and private bases frozen", | |
| "merge": "C' = C + (alpha/r) B_lora A_lora, then W = C' @ basis", | |
| "mean_rel_delta_coefficient": 0.023527406085382348 | |
| }, | |
| "param_fraction_is_upper_bound": false, | |
| "rel_delta_coefficient": { | |
| "0.down_proj": 0.01414614636451006, | |
| "0.gate_proj": 0.06271173059940338, | |
| "0.up_proj": 0.047500111162662506, | |
| "0.k_proj": 0.028756480664014816, | |
| "0.o_proj": 0.012572512030601501, | |
| "0.q_proj": 0.030865762382745743, | |
| "0.v_proj": 0.01895243488252163, | |
| "1.down_proj": 0.00678357295691967, | |
| "1.gate_proj": 0.04598936438560486, | |
| "1.up_proj": 0.040169041603803635, | |
| "1.k_proj": 0.014593809843063354, | |
| "1.o_proj": 0.01309526152908802, | |
| "1.q_proj": 0.013989835046231747, | |
| "1.v_proj": 0.014499799348413944, | |
| "2.down_proj": 0.013051525689661503, | |
| "2.gate_proj": 0.040834009647369385, | |
| "2.up_proj": 0.03902892395853996, | |
| "2.k_proj": 0.01710350811481476, | |
| "2.o_proj": 0.015589666552841663, | |
| "2.q_proj": 0.015468768775463104, | |
| "2.v_proj": 0.021661514416337013, | |
| "3.down_proj": 0.013177504763007164, | |
| "3.gate_proj": 0.03536074981093407, | |
| "3.up_proj": 0.03635535389184952, | |
| "3.k_proj": 0.016993820667266846, | |
| "3.o_proj": 0.01536262035369873, | |
| "3.q_proj": 0.01714141294360161, | |
| "3.v_proj": 0.02327631041407585, | |
| "4.down_proj": 0.012231411412358284, | |
| "4.gate_proj": 0.03274000436067581, | |
| "4.up_proj": 0.03400220721960068, | |
| "4.k_proj": 0.018720826134085655, | |
| "4.o_proj": 0.015074113383889198, | |
| "4.q_proj": 0.01642892323434353, | |
| "4.v_proj": 0.020959772169589996, | |
| "5.down_proj": 0.01268017292022705, | |
| "5.gate_proj": 0.03295743092894554, | |
| "5.up_proj": 0.032916516065597534, | |
| "5.k_proj": 0.016637030988931656, | |
| "5.o_proj": 0.014721281826496124, | |
| "5.q_proj": 0.015442839823663235, | |
| "5.v_proj": 0.020913822576403618, | |
| "6.down_proj": 0.01286441832780838, | |
| "6.gate_proj": 0.03249712288379669, | |
| "6.up_proj": 0.02966822311282158, | |
| "6.k_proj": 0.017795151099562645, | |
| "6.o_proj": 0.014706257730722427, | |
| "6.q_proj": 0.015420209616422653, | |
| "6.v_proj": 0.019306669011712074, | |
| "7.down_proj": 0.012474670074880123, | |
| "7.gate_proj": 0.034134846180677414, | |
| "7.up_proj": 0.03099922463297844, | |
| "7.k_proj": 0.016013693064451218, | |
| "7.o_proj": 0.014082911424338818, | |
| "7.q_proj": 0.014865195378661156, | |
| "7.v_proj": 0.020185057073831558, | |
| "8.down_proj": 0.012424257583916187, | |
| "8.gate_proj": 0.0344713032245636, | |
| "8.up_proj": 0.03093084692955017, | |
| "8.k_proj": 0.01666663959622383, | |
| "8.o_proj": 0.01538773626089096, | |
| "8.q_proj": 0.015956632792949677, | |
| "8.v_proj": 0.018667476251721382, | |
| "9.down_proj": 0.012787723913788795, | |
| "9.gate_proj": 0.030635781586170197, | |
| "9.up_proj": 0.03066754899919033, | |
| "9.k_proj": 0.01625077985227108, | |
| "9.o_proj": 0.016457993537187576, | |
| "9.q_proj": 0.015760647132992744, | |
| "9.v_proj": 0.01802702806890011, | |
| "10.down_proj": 0.013270135037600994, | |
| "10.gate_proj": 0.03306710347533226, | |
| "10.up_proj": 0.031808458268642426, | |
| "10.k_proj": 0.016889700666069984, | |
| "10.o_proj": 0.014808307401835918, | |
| "10.q_proj": 0.016796549782156944, | |
| "10.v_proj": 0.016769904643297195, | |
| "11.down_proj": 0.01278386078774929, | |
| "11.gate_proj": 0.032750505954027176, | |
| "11.up_proj": 0.03186879679560661, | |
| "11.k_proj": 0.018679214641451836, | |
| "11.o_proj": 0.01585085503757, | |
| "11.q_proj": 0.017030365765094757, | |
| "11.v_proj": 0.016336703673005104, | |
| "12.down_proj": 0.013383959420025349, | |
| "12.gate_proj": 0.03204452618956566, | |
| "12.up_proj": 0.031297486275434494, | |
| "12.k_proj": 0.018261687830090523, | |
| "12.o_proj": 0.015123218297958374, | |
| "12.q_proj": 0.018485387787222862, | |
| "12.v_proj": 0.01751946099102497, | |
| "13.down_proj": 0.01318136602640152, | |
| "13.gate_proj": 0.033618710935115814, | |
| "13.up_proj": 0.03187503293156624, | |
| "13.k_proj": 0.019344637170433998, | |
| "13.o_proj": 0.01476432103663683, | |
| "13.q_proj": 0.01824866235256195, | |
| "13.v_proj": 0.018521904945373535, | |
| "14.down_proj": 0.013778213411569595, | |
| "14.gate_proj": 0.0355733186006546, | |
| "14.up_proj": 0.03251698985695839, | |
| "14.k_proj": 0.019346466287970543, | |
| "14.o_proj": 0.015444910153746605, | |
| "14.q_proj": 0.01895812153816223, | |
| "14.v_proj": 0.01812027022242546, | |
| "15.down_proj": 0.013836321420967579, | |
| "15.gate_proj": 0.033386047929525375, | |
| "15.up_proj": 0.03200124576687813, | |
| "15.k_proj": 0.020122358575463295, | |
| "15.o_proj": 0.016066821292042732, | |
| "15.q_proj": 0.019362781196832657, | |
| "15.v_proj": 0.017482535913586617, | |
| "16.down_proj": 0.014055527746677399, | |
| "16.gate_proj": 0.03231862932443619, | |
| "16.up_proj": 0.03309623897075653, | |
| "16.k_proj": 0.022296005859971046, | |
| "16.o_proj": 0.015972847118973732, | |
| "16.q_proj": 0.02107972837984562, | |
| "16.v_proj": 0.022093387320637703, | |
| "17.down_proj": 0.01484985463321209, | |
| "17.gate_proj": 0.03461271524429321, | |
| "17.up_proj": 0.035233646631240845, | |
| "17.k_proj": 0.02109636552631855, | |
| "17.o_proj": 0.016126474365592003, | |
| "17.q_proj": 0.0208114143460989, | |
| "17.v_proj": 0.02159978821873665, | |
| "18.down_proj": 0.015112471766769886, | |
| "18.gate_proj": 0.03760548308491707, | |
| "18.up_proj": 0.03554264083504677, | |
| "18.k_proj": 0.022595606744289398, | |
| "18.o_proj": 0.01604403182864189, | |
| "18.q_proj": 0.020856209099292755, | |
| "18.v_proj": 0.02414970099925995, | |
| "19.down_proj": 0.01540366467088461, | |
| "19.gate_proj": 0.03634091094136238, | |
| "19.up_proj": 0.0359489843249321, | |
| "19.k_proj": 0.02256142720580101, | |
| "19.o_proj": 0.017474617809057236, | |
| "19.q_proj": 0.02237251214683056, | |
| "19.v_proj": 0.022501647472381592, | |
| "20.down_proj": 0.016096290200948715, | |
| "20.gate_proj": 0.0366799496114254, | |
| "20.up_proj": 0.037445854395627975, | |
| "20.k_proj": 0.02196851558983326, | |
| "20.o_proj": 0.017981523647904396, | |
| "20.q_proj": 0.023315971717238426, | |
| "20.v_proj": 0.023550692945718765, | |
| "21.down_proj": 0.015590106137096882, | |
| "21.gate_proj": 0.038024019449949265, | |
| "21.up_proj": 0.039658188819885254, | |
| "21.k_proj": 0.022505933418869972, | |
| "21.o_proj": 0.017578478902578354, | |
| "21.q_proj": 0.021982474252581596, | |
| "21.v_proj": 0.024720672518014908, | |
| "22.down_proj": 0.015707720071077347, | |
| "22.gate_proj": 0.03906581178307533, | |
| "22.up_proj": 0.04101793095469475, | |
| "22.k_proj": 0.022656260058283806, | |
| "22.o_proj": 0.01720021851360798, | |
| "22.q_proj": 0.0214886162430048, | |
| "22.v_proj": 0.02467689849436283, | |
| "23.down_proj": 0.015716182067990303, | |
| "23.gate_proj": 0.041364703327417374, | |
| "23.up_proj": 0.041586700826883316, | |
| "23.k_proj": 0.021899448707699776, | |
| "23.o_proj": 0.016503864899277687, | |
| "23.q_proj": 0.022801637649536133, | |
| "23.v_proj": 0.0233846977353096, | |
| "24.down_proj": 0.01567365787923336, | |
| "24.gate_proj": 0.04018592834472656, | |
| "24.up_proj": 0.04268543794751167, | |
| "24.k_proj": 0.022960152477025986, | |
| "24.o_proj": 0.015794076025485992, | |
| "24.q_proj": 0.024518316611647606, | |
| "24.v_proj": 0.024992572143673897, | |
| "25.down_proj": 0.01578501984477043, | |
| "25.gate_proj": 0.0379466786980629, | |
| "25.up_proj": 0.04037538915872574, | |
| "25.k_proj": 0.024068159982562065, | |
| "25.o_proj": 0.017214398831129074, | |
| "25.q_proj": 0.02083326317369938, | |
| "25.v_proj": 0.026051005348563194, | |
| "26.down_proj": 0.014912595972418785, | |
| "26.gate_proj": 0.039050161838531494, | |
| "26.up_proj": 0.03992636501789093, | |
| "26.k_proj": 0.02621399611234665, | |
| "26.o_proj": 0.014508343301713467, | |
| "26.q_proj": 0.02345091477036476, | |
| "26.v_proj": 0.02503756619989872, | |
| "27.down_proj": 0.015143311582505703, | |
| "27.gate_proj": 0.038240697234869, | |
| "27.up_proj": 0.03937762975692749, | |
| "27.k_proj": 0.021424096077680588, | |
| "27.o_proj": 0.01971602812409401, | |
| "27.q_proj": 0.022661816328763962, | |
| "27.v_proj": 0.02328498102724552, | |
| "28.down_proj": 0.01482220645993948, | |
| "28.gate_proj": 0.04094494506716728, | |
| "28.up_proj": 0.03843633458018303, | |
| "28.k_proj": 0.02151227742433548, | |
| "28.o_proj": 0.01504539605230093, | |
| "28.q_proj": 0.021972497925162315, | |
| "28.v_proj": 0.023144179955124855, | |
| "29.down_proj": 0.015450587496161461, | |
| "29.gate_proj": 0.03961341828107834, | |
| "29.up_proj": 0.039220333099365234, | |
| "29.k_proj": 0.02610217034816742, | |
| "29.o_proj": 0.014716818928718567, | |
| "29.q_proj": 0.0211815033107996, | |
| "29.v_proj": 0.022618116810917854, | |
| "30.down_proj": 0.012776860035955906, | |
| "30.gate_proj": 0.038771044462919235, | |
| "30.up_proj": 0.03742718696594238, | |
| "30.k_proj": 0.0264947060495615, | |
| "30.o_proj": 0.015689056366682053, | |
| "30.q_proj": 0.02465638890862465, | |
| "30.v_proj": 0.018169045448303223, | |
| "31.down_proj": 0.010587853379547596, | |
| "31.gate_proj": 0.03626640886068344, | |
| "31.up_proj": 0.034017015248537064, | |
| "31.k_proj": 0.03141297027468681, | |
| "31.o_proj": 0.01466062106192112, | |
| "31.q_proj": 0.0292032640427351, | |
| "31.v_proj": 0.01933012343943119 | |
| } | |
| } |