Text Generation
Transformers
Safetensors
English
gemma4
uncensored
abliterated
forensics
thinking
reasoning
2026
Instructions to use DreamFast/Gemma4-e4b-abliterlitics with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use DreamFast/Gemma4-e4b-abliterlitics with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="DreamFast/Gemma4-e4b-abliterlitics")# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("DreamFast/Gemma4-e4b-abliterlitics", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use DreamFast/Gemma4-e4b-abliterlitics with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "DreamFast/Gemma4-e4b-abliterlitics" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "DreamFast/Gemma4-e4b-abliterlitics", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/DreamFast/Gemma4-e4b-abliterlitics
- SGLang
How to use DreamFast/Gemma4-e4b-abliterlitics 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 "DreamFast/Gemma4-e4b-abliterlitics" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "DreamFast/Gemma4-e4b-abliterlitics", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'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 "DreamFast/Gemma4-e4b-abliterlitics" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "DreamFast/Gemma4-e4b-abliterlitics", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use DreamFast/Gemma4-e4b-abliterlitics with Docker Model Runner:
docker model run hf.co/DreamFast/Gemma4-e4b-abliterlitics
| { | |
| "metadata": { | |
| "tool": "abliterlitics", | |
| "version": "1.0.0", | |
| "results_version": 1, | |
| "comparison_name": "gemma4-e4b", | |
| "variant": "coder3101", | |
| "architecture": "gemma4", | |
| "timestamp": "2026-06-27T11:53:28.815513+00:00" | |
| }, | |
| "results_version": 1, | |
| "label": "Coder3101 Heretic", | |
| "architecture": { | |
| "family": "gemma4", | |
| "layers": 42, | |
| "total_keys": 2130, | |
| "experts": 0 | |
| }, | |
| "scope": { | |
| "total_tensors": 719, | |
| "changed_tensors": 21, | |
| "changed_pct": 2.920723226703755, | |
| "total_params_changed": 131072000, | |
| "param_edit_density": 1.743426395890155 | |
| }, | |
| "magnitude": { | |
| "edit_norm_mean": 2.078418173960277, | |
| "edit_norm_median": 2.122333526611328, | |
| "edit_norm_p25": 1.84255850315094, | |
| "edit_norm_p75": 2.3862524032592773, | |
| "edit_norm_p95": 2.586658239364624, | |
| "relative_edit_mean": 0.03599294724184891, | |
| "relative_edit_median": 0.03732164958191203 | |
| }, | |
| "targeting": { | |
| "tensor_types": { | |
| "self_attn.o_proj.weight": 21 | |
| }, | |
| "categories": { | |
| "attention": 21 | |
| }, | |
| "layers_modified": 21, | |
| "layer_coverage_pct": 50.0 | |
| }, | |
| "layer_profile": { | |
| "mean_edits_per_layer": 1.0, | |
| "max_edits_layer": 1, | |
| "layer_entropy": 81.45509765846309, | |
| "early_layer_pct": 0.0, | |
| "mid_layer_pct": 33.33333333333333, | |
| "late_layer_pct": 66.66666666666666 | |
| }, | |
| "per_key": [ | |
| { | |
| "key": "model.language_model.embed_tokens.weight", | |
| "type": "embed_tokens.weight", | |
| "layer": null, | |
| "category": "embedding", | |
| "edit_norm": 0.0, | |
| "base_norm": 539.5653686523438, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 671088640 | |
| }, | |
| { | |
| "key": "model.language_model.embed_tokens_per_layer.weight", | |
| "type": "embed_tokens_per_layer.weight", | |
| "layer": null, | |
| "category": "embedding", | |
| "edit_norm": 0.0, | |
| "base_norm": 2139.47021484375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2818572288 | |
| }, | |
| { | |
| "key": "model.language_model.layers.0.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 0, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 562.3897705078125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.0.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 0, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.06103515625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.0.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 0, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 106.39189910888672, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.0.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 0, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 131.32278442382812, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.0.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 0, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 134.34947204589844, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.0.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 0, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.1104313135147095, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.0.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 0, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 56.417808532714844, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.0.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 0, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 363.11517333984375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.0.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 0, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 397.9847106933594, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.0.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 0, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 283.6213073730469, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.0.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 0, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 334.4049072265625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.0.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 0, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.03125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.0.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 0, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 31.77863311767578, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.0.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 0, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 63.56478500366211, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.0.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 0, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.75, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.0.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 0, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 63.10002517700195, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.0.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 0, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 31.6990966796875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.1.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 1, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 719.3878173828125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.1.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 1, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.16015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.1.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 1, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 108.31302642822266, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.1.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 1, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 132.3343963623047, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.1.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 1, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 136.81423950195312, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.1.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 1, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 3.1151113510131836, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.1.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 1, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 53.24144744873047, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.1.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 1, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 216.82435607910156, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.1.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 1, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 61.07600784301758, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.1.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 1, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.609729766845703, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.1.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 1, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 2292.366943359375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.1.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 1, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.078125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.1.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 1, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 30.38418960571289, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.1.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 1, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 59.78000259399414, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.1.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 1, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.4375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.1.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 1, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 61.95423126220703, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.1.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 1, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 32.20498275756836, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.10.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 10, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 1547.0220947265625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.10.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 10, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.62109375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.10.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 10, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 98.68649291992188, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.10.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 10, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 137.78512573242188, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.10.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 10, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 138.1079864501953, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.10.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 10, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 13.980788230895996, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.10.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 10, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.16932678222656, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.10.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 10, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 34.16542434692383, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.10.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 10, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 65.27615356445312, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.10.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 10, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 34.191768646240234, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.10.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 10, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 1192.8363037109375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.10.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 10, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.953125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.10.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 10, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 31.176658630371094, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.10.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 10, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 65.95577239990234, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.10.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 10, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.10.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 10, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 61.56078338623047, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.10.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 10, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 31.098133087158203, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.11.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 11, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 2193.435546875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.11.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 11, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.51953125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.11.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 11, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 100.94623565673828, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.11.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 11, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 132.86460876464844, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.11.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 11, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 132.35227966308594, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.11.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 11, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 14.63333511352539, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.11.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 11, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 55.30987548828125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.11.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 11, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 43.550437927246094, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.11.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 11, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 77.1336669921875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.11.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 11, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 68.26229095458984, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.11.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 11, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 1687.37646484375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.11.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 11, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.4584077596664429, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 512 | |
| }, | |
| { | |
| "key": "model.language_model.layers.11.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 11, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 40.70024108886719, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2621440 | |
| }, | |
| { | |
| "key": "model.language_model.layers.11.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 11, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 71.12318420410156, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 10485760 | |
| }, | |
| { | |
| "key": "model.language_model.layers.11.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 11, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 22.008699417114258, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 512 | |
| }, | |
| { | |
| "key": "model.language_model.layers.11.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 11, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 81.70069122314453, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 10485760 | |
| }, | |
| { | |
| "key": "model.language_model.layers.11.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 11, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 40.47544860839844, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2621440 | |
| }, | |
| { | |
| "key": "model.language_model.layers.12.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 12, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 1976.65234375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.12.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 12, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.74609375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.12.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 12, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 101.75801086425781, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.12.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 12, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 141.16261291503906, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.12.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 12, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 142.0087127685547, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.12.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 12, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.604171752929688, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.12.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 12, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 53.95722579956055, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.12.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 12, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 36.50940704345703, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.12.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 12, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 51.913665771484375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.12.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 12, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.453725814819336, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.12.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 12, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 1928.5867919921875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.12.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 12, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.921875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.12.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 12, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 29.890857696533203, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.12.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 12, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 61.914859771728516, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.12.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 12, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.12.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 12, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 59.62895965576172, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.12.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 12, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 29.807828903198242, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.13.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 13, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 2129.828369140625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.13.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 13, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.7109375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.13.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 13, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 104.60979461669922, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.13.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 13, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 137.23255920410156, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.13.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 13, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 137.30419921875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.13.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 13, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 14.983562469482422, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.13.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 13, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.9785041809082, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.13.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 13, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 24.908485412597656, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.13.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 13, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 68.48486328125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.13.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 13, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 32.42045593261719, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.13.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 13, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 873.1041870117188, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.13.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 13, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.953125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.13.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 13, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 28.930240631103516, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.13.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 13, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 63.7784309387207, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.13.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 13, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.13.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 13, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 56.93617248535156, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.13.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 13, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 28.80209732055664, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.14.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 14, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 1298.9908447265625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.14.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 14, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.671875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.14.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 14, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 97.69435119628906, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.14.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 14, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 130.90371704101562, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.14.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 14, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 130.38755798339844, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.14.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 14, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 13.424017906188965, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.14.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 14, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.88349151611328, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.14.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 14, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 32.06250762939453, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.14.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 14, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 87.68759155273438, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.14.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 14, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 14.200285911560059, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.14.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 14, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 244.29563903808594, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.14.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 14, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.9375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.14.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 14, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 28.639606475830078, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.14.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 14, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 63.928192138671875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.14.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 14, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.5, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.14.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 14, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 56.36750793457031, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.14.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 14, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 28.187314987182617, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.15.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 15, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 660.2973022460938, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.15.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 15, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.5390625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.15.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 15, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 89.46987915039062, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.15.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 15, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 129.58055114746094, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.15.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 15, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 129.13665771484375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.15.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 15, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 11.995497703552246, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.15.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 15, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 50.15523910522461, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.15.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 15, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 50.034461975097656, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.15.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 15, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 106.54208374023438, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.15.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 15, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 61.7928352355957, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.15.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 15, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 147.4600372314453, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.15.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 15, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.03125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.15.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 15, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 27.489768981933594, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.15.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 15, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 58.18947982788086, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.15.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 15, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.75, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.15.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 15, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.83394241333008, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.15.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 15, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 26.59971046447754, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.16.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 16, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 764.9962158203125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.16.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 16, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.63671875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.16.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 16, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 83.39749145507812, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.16.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 16, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 118.96552276611328, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.16.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 16, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 118.37474060058594, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.16.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 16, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 13.103943824768066, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.16.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 16, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 51.46934127807617, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.16.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 16, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 41.71900939941406, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.16.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 16, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 91.3440933227539, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.16.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 16, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 29.92510986328125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.16.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 16, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 60.94634246826172, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.16.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 16, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.16.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 16, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 26.9898738861084, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.16.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 16, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 56.821624755859375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.16.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 16, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.9375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.16.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 16, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.77863693237305, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.16.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 16, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 26.24455451965332, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.17.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 17, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 375.3606872558594, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.17.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 17, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.423828125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.17.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 17, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 82.90483856201172, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.17.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 17, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 115.16876983642578, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.17.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 17, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 116.63542175292969, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.17.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 17, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 9.333442687988281, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.17.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 17, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 50.74952697753906, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.17.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 17, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 71.1190185546875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.17.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 17, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 151.1227264404297, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.17.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 17, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 63.36868667602539, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.17.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 17, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 50.456329345703125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.17.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 17, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.408689260482788, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 512 | |
| }, | |
| { | |
| "key": "model.language_model.layers.17.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 17, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 35.82019805908203, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2621440 | |
| }, | |
| { | |
| "key": "model.language_model.layers.17.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 17, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 63.41203689575195, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 10485760 | |
| }, | |
| { | |
| "key": "model.language_model.layers.17.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 17, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 22.8041934967041, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 512 | |
| }, | |
| { | |
| "key": "model.language_model.layers.17.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 17, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 74.56464385986328, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 10485760 | |
| }, | |
| { | |
| "key": "model.language_model.layers.17.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 17, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 35.880550384521484, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2621440 | |
| }, | |
| { | |
| "key": "model.language_model.layers.18.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 18, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 494.3365783691406, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.18.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 18, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.54296875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.18.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 18, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 75.4670639038086, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.18.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 18, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 117.05247497558594, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.18.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 18, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 118.57843780517578, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.18.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 18, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 11.43996810913086, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.18.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 18, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.02887725830078, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.18.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 18, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 52.17005920410156, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.18.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 18, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 117.50435638427734, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.18.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 18, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 14.165892601013184, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.18.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 18, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 77.17996978759766, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.18.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 18, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.9453125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.18.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 18, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 25.950281143188477, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.18.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 18, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 55.12913513183594, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.18.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 18, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.5, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.18.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 18, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.7093620300293, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.18.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 18, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 26.021678924560547, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.19.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 19, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 394.1702880859375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.19.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 19, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.51171875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.19.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 19, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 73.48778533935547, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.19.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 19, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 112.24176025390625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.19.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 19, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 113.67285919189453, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.19.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 19, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 10.785642623901367, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.19.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 19, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 52.46030807495117, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.19.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 19, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 51.172054290771484, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.19.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 19, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 127.5758056640625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.19.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 19, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 9.515192985534668, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.19.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 19, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 113.52222442626953, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.19.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 19, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.984375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.19.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 19, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 26.394371032714844, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.19.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 19, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 57.19216537475586, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.19.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 19, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.19.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 19, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.30415344238281, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.19.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 19, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 26.505474090576172, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.2.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 2, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 2320.781494140625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.2.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 2, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.83984375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.2.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 2, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 105.21880340576172, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.2.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 2, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 141.4124755859375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.2.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 2, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 141.7940673828125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.2.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 2, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.456470489501953, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.2.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 2, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 53.56603240966797, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.2.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 2, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 20.98313331604004, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.2.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 2, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 34.59356689453125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.2.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 2, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 5.0112714767456055, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.2.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 2, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 1515.034912109375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.2.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 2, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.109375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.2.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 2, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 28.582338333129883, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.2.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 2, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 67.74276733398438, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.2.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 2, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.1875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.2.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 2, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 56.966156005859375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.2.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 2, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 29.444568634033203, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.20.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 20, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 338.66387939453125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.20.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 20, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.431640625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.20.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 20, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 73.04698944091797, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.20.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 20, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 111.29114532470703, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.20.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 20, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 111.69947052001953, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.20.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 20, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 9.136544227600098, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.20.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 20, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 50.18399429321289, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.20.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 20, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 59.49087142944336, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.20.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 20, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 153.1579132080078, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.20.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 20, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 41.64082717895508, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.20.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 20, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 177.44410705566406, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.20.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 20, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.0, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.20.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 20, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 26.20272445678711, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.20.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 20, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 57.4205322265625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.20.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 20, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.0, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.20.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 20, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 53.92119598388672, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.20.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 20, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 26.003355026245117, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.21.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 21, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 374.95904541015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.21.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 21, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.4609375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.21.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 21, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 72.85941314697266, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.21.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 21, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 113.86516571044922, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.21.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 21, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 113.813720703125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.21.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 21, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 9.560681343078613, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.21.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 21, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 52.329830169677734, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.21.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 21, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 49.46298599243164, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.21.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 21, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 147.39613342285156, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.21.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 21, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 21.354665756225586, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.21.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 21, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 395.9770812988281, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.21.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 21, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.046875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.21.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 21, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.130126953125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.21.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 21, | |
| "category": "attention", | |
| "edit_norm": 1.492377758026123, | |
| "base_norm": 52.244346618652344, | |
| "rel_norm": 0.028565344474866347, | |
| "changed": true, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.21.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 21, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.6875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.21.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 21, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 47.91935348510742, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.21.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 21, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.27240562438965, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.22.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 22, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 515.6870727539062, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.22.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 22, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.1572265625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.22.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 22, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 72.08293914794922, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.22.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 22, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 106.51625061035156, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.22.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 22, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 106.04823303222656, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.22.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 22, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 4.582650184631348, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.22.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 22, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 51.54432678222656, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.22.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 22, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 45.32465362548828, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.22.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 22, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 393.4920654296875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.22.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 22, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 162.01419067382812, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.22.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 22, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 392.6501159667969, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.22.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 22, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.22.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 22, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 24.960397720336914, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.22.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 22, | |
| "category": "attention", | |
| "edit_norm": 2.1177456378936768, | |
| "base_norm": 56.74308776855469, | |
| "rel_norm": 0.03732164958191203, | |
| "changed": true, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.22.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 22, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.22.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 22, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.099544525146484, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.22.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 22, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 25.08211898803711, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.23.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 23, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 642.0862426757812, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.23.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 23, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.0654296875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.23.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 23, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 74.24793243408203, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.23.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 23, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 118.47179412841797, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.23.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 23, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 121.57816314697266, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.23.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 23, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.226837396621704, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.23.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 23, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 53.387550354003906, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.23.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 23, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 284.7088623046875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.23.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 23, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 1351.4755859375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.23.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 23, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 434.1996154785156, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.23.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 23, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 191.7904052734375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.23.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 23, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.3810679912567139, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 512 | |
| }, | |
| { | |
| "key": "model.language_model.layers.23.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 23, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 37.095096588134766, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2621440 | |
| }, | |
| { | |
| "key": "model.language_model.layers.23.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 23, | |
| "category": "attention", | |
| "edit_norm": 1.7904515266418457, | |
| "base_norm": 62.50578308105469, | |
| "rel_norm": 0.02864457396398136, | |
| "changed": true, | |
| "numel": 10485760 | |
| }, | |
| { | |
| "key": "model.language_model.layers.23.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 23, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.157747268676758, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 512 | |
| }, | |
| { | |
| "key": "model.language_model.layers.23.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 23, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 79.68150329589844, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 10485760 | |
| }, | |
| { | |
| "key": "model.language_model.layers.23.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 23, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 37.08687973022461, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2621440 | |
| }, | |
| { | |
| "key": "model.language_model.layers.24.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 24, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 575.7376098632812, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.24.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 24, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.50390625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.24.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 24, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 85.16482543945312, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.24.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 24, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 121.41227722167969, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.24.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 24, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 125.4738998413086, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.24.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 24, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 9.340764045715332, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.24.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 24, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 49.77634048461914, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.24.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 24, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 63.883026123046875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.24.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 24, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 144.25477600097656, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.24.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 24, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 49.134979248046875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.24.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 24, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 105.42514038085938, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.24.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 24, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.24.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 24, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.254680633544922, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.24.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 24, | |
| "category": "attention", | |
| "edit_norm": 2.4156219959259033, | |
| "base_norm": 54.16752243041992, | |
| "rel_norm": 0.044595393836387005, | |
| "changed": true, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.24.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 24, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.24.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 24, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 55.34669494628906, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.24.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 24, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.254491806030273, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.25.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 25, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 422.9901428222656, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.25.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 25, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.71484375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.25.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 25, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 83.83642578125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.25.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 25, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 121.14268493652344, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.25.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 25, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 124.93534851074219, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.25.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 25, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 14.066128730773926, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.25.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 25, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 50.683448791503906, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.25.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 25, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 38.610557556152344, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.25.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 25, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 79.81321716308594, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.25.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 25, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 28.299955368041992, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.25.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 25, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 28.58111000061035, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.25.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 25, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.25.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 25, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 22.87788200378418, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.25.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 25, | |
| "category": "attention", | |
| "edit_norm": 2.523261308670044, | |
| "base_norm": 58.29996109008789, | |
| "rel_norm": 0.043280668828766106, | |
| "changed": true, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.25.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 25, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.25.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 25, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.85015869140625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.25.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 25, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 22.87778091430664, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.26.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 26, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 324.03106689453125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.26.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 26, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.71484375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.26.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 26, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 83.2928466796875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.26.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 26, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 113.37691497802734, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.26.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 26, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 116.79265594482422, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.26.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 26, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 14.083134651184082, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.26.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 26, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 49.41382598876953, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.26.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 26, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 33.47078323364258, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.26.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 26, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 79.9291763305664, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.26.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 26, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 32.05538558959961, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.26.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 26, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 32.914466857910156, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.26.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 26, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.26.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 26, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.7338924407959, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.26.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 26, | |
| "category": "attention", | |
| "edit_norm": 1.9957588911056519, | |
| "base_norm": 58.45262145996094, | |
| "rel_norm": 0.03414318881271584, | |
| "changed": true, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.26.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 26, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.26.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 26, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 55.5416259765625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.26.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 26, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.73393440246582, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.27.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 27, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 306.5792236328125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.27.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 27, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.69140625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.27.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 27, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 80.03752136230469, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.27.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 27, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 120.77427673339844, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.27.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 27, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 124.40129089355469, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.27.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 27, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 13.151424407958984, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.27.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 27, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 49.37179183959961, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.27.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 27, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 34.83408737182617, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.27.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 27, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 87.82120513916016, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.27.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 27, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 28.082853317260742, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.27.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 27, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 38.691646575927734, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.27.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 27, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.27.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 27, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 22.959409713745117, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.27.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 27, | |
| "category": "attention", | |
| "edit_norm": 2.1132750511169434, | |
| "base_norm": 57.23421096801758, | |
| "rel_norm": 0.0369232844373069, | |
| "changed": true, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.27.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 27, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.27.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 27, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 52.71851348876953, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.27.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 27, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 22.95948600769043, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.28.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 28, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 314.63037109375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.28.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 28, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.61328125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.28.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 28, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 77.897705078125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.28.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 28, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 116.3714828491211, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.28.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 28, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 118.85167694091797, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.28.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 28, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 11.97512435913086, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.28.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 28, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 49.099830627441406, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.28.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 28, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 37.251399993896484, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.28.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 28, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 102.69879150390625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.28.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 28, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 46.795623779296875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.28.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 28, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 51.12250518798828, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.28.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 28, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.28.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 28, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.3483943939209, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.28.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 28, | |
| "category": "attention", | |
| "edit_norm": 1.6939581632614136, | |
| "base_norm": 54.24738311767578, | |
| "rel_norm": 0.031226541556609394, | |
| "changed": true, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.28.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 28, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.28.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 28, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.602420806884766, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.28.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 28, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.34839630126953, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.29.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 29, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 249.12576293945312, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.29.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 29, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.419921875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.29.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 29, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 78.30254364013672, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.29.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 29, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 122.4095458984375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.29.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 29, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 121.03590393066406, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.29.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 29, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 8.439972877502441, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.29.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 29, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.03239822387695, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.29.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 29, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 51.67462158203125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.29.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 29, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 145.17466735839844, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.29.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 29, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.89714813232422, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.29.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 29, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 37.648231506347656, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.29.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 29, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.3810679912567139, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 512 | |
| }, | |
| { | |
| "key": "model.language_model.layers.29.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 29, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 32.213905334472656, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2621440 | |
| }, | |
| { | |
| "key": "model.language_model.layers.29.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 29, | |
| "category": "attention", | |
| "edit_norm": 2.586658239364624, | |
| "base_norm": 59.80091857910156, | |
| "rel_norm": 0.0432544900784948, | |
| "changed": true, | |
| "numel": 10485760 | |
| }, | |
| { | |
| "key": "model.language_model.layers.29.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 29, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.157747268676758, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 512 | |
| }, | |
| { | |
| "key": "model.language_model.layers.29.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 29, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 72.57662200927734, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 10485760 | |
| }, | |
| { | |
| "key": "model.language_model.layers.29.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 29, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 32.21397018432617, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2621440 | |
| }, | |
| { | |
| "key": "model.language_model.layers.3.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 3, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 2028.5081787109375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.3.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 3, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.7109375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.3.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 3, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 104.7504653930664, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.3.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 3, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 132.50338745117188, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.3.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 3, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 133.5342559814453, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.3.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 3, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 14.070670127868652, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.3.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 3, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 52.58344650268555, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.3.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 3, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 29.851856231689453, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.3.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 3, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 49.2585563659668, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.3.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 3, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 4.856971740722656, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.3.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 3, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 1691.3837890625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.3.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 3, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.9609375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.3.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 3, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 30.759769439697266, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.3.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 3, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 63.57655715942383, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.3.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 3, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.3.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 3, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 62.06000900268555, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.3.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 3, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 31.983732223510742, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.30.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 30, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 275.03326416015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.30.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 30, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.7109375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.30.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 30, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 87.99722290039062, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.30.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 30, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 123.37210083007812, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.30.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 30, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 122.2601318359375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.30.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 30, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 14.17929744720459, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.30.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 30, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 49.06552505493164, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.30.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 30, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 38.25714111328125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.30.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 30, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 79.1627197265625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.30.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 30, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 11.472908020019531, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.30.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 30, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.75308609008789, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.30.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 30, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.30.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 30, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 22.21617317199707, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.30.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 30, | |
| "category": "attention", | |
| "edit_norm": 2.3862524032592773, | |
| "base_norm": 55.84449005126953, | |
| "rel_norm": 0.042730310565438315, | |
| "changed": true, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.30.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 30, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.30.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 30, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 51.364471435546875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.30.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 30, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 22.216140747070312, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.31.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 31, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 334.6661376953125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.31.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 31, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.81640625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.31.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 31, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 88.03897094726562, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.31.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 31, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 120.36177825927734, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.31.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 31, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 120.38553619384766, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.31.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 31, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.187379837036133, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.31.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 31, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 51.449588775634766, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.31.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 31, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 31.8814754486084, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.31.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 31, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 58.48835754394531, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.31.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 31, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 14.449666023254395, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.31.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 31, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 46.89055633544922, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.31.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 31, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.31.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 31, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.459720611572266, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.31.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 31, | |
| "category": "attention", | |
| "edit_norm": 3.4110827445983887, | |
| "base_norm": 56.45244216918945, | |
| "rel_norm": 0.06042400671303615, | |
| "changed": true, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.31.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 31, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.31.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 31, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 55.74137496948242, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.31.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 31, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.459741592407227, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.32.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 32, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 315.878662109375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.32.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 32, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.8515625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.32.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 32, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 87.92033386230469, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.32.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 32, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 129.4443359375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.32.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 32, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 128.67242431640625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.32.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 32, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.886133193969727, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.32.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 32, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.05058670043945, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.32.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 32, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 24.39735221862793, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.32.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 32, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.387451171875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.32.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 32, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 12.636390686035156, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.32.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 32, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 57.66130828857422, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.32.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 32, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.32.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 32, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.75921058654785, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.32.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 32, | |
| "category": "attention", | |
| "edit_norm": 2.2904324531555176, | |
| "base_norm": 58.05116653442383, | |
| "rel_norm": 0.03945540787362665, | |
| "changed": true, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.32.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 32, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.32.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 32, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.3385009765625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.32.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 32, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.759166717529297, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.33.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 33, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 248.97097778320312, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.33.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 33, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.82421875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.33.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 33, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 93.35466766357422, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.33.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 33, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 124.56626892089844, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.33.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 33, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 124.0318832397461, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.33.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 33, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.25665855407715, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.33.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 33, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 51.95609664916992, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.33.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 33, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 32.4841194152832, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.33.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 33, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 57.421974182128906, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.33.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 33, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 14.143991470336914, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.33.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 33, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 58.04665756225586, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.33.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 33, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.33.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 33, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.140031814575195, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.33.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 33, | |
| "category": "attention", | |
| "edit_norm": 2.122333526611328, | |
| "base_norm": 56.562347412109375, | |
| "rel_norm": 0.037522019925165975, | |
| "changed": true, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.33.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 33, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.33.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 33, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 53.74324417114258, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.33.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 33, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.140012741088867, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.34.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 34, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 304.1881103515625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.34.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 34, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.8359375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.34.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 34, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 91.57036590576172, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.34.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 34, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 129.44183349609375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.34.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 34, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 128.364013671875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.34.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 34, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.66167449951172, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.34.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 34, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.073089599609375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.34.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 34, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 22.000741958618164, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.34.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 34, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 59.34286880493164, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.34.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 34, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 12.755073547363281, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.34.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 34, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 52.420108795166016, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.34.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 34, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.34.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 34, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.332046508789062, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.34.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 34, | |
| "category": "attention", | |
| "edit_norm": 2.3641724586486816, | |
| "base_norm": 58.65534973144531, | |
| "rel_norm": 0.04030616933447831, | |
| "changed": true, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.34.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 34, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.34.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 34, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 53.4749641418457, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.34.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 34, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.332014083862305, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.35.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 35, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 294.75921630859375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.35.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 35, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.8359375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.35.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 35, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 88.97764587402344, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.35.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 35, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 128.37403869628906, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.35.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 35, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 129.5680389404297, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.35.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 35, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.661291122436523, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.35.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 35, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 49.28533935546875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.35.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 35, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 31.467853546142578, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.35.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 35, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 55.94285583496094, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.35.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 35, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 13.204005241394043, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.35.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 35, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 37.50951385498047, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.35.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 35, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.3810679912567139, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 512 | |
| }, | |
| { | |
| "key": "model.language_model.layers.35.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 35, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 31.94091033935547, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2621440 | |
| }, | |
| { | |
| "key": "model.language_model.layers.35.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 35, | |
| "category": "attention", | |
| "edit_norm": 1.3407988548278809, | |
| "base_norm": 65.68130493164062, | |
| "rel_norm": 0.02041370609526332, | |
| "changed": true, | |
| "numel": 10485760 | |
| }, | |
| { | |
| "key": "model.language_model.layers.35.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 35, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.157747268676758, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 512 | |
| }, | |
| { | |
| "key": "model.language_model.layers.35.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 35, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 74.2364501953125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 10485760 | |
| }, | |
| { | |
| "key": "model.language_model.layers.35.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 35, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 31.941055297851562, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2621440 | |
| }, | |
| { | |
| "key": "model.language_model.layers.36.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 36, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 380.3974914550781, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.36.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 36, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.85546875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.36.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 36, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 89.95733642578125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.36.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 36, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 129.58187866210938, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.36.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 36, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 131.81524658203125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.36.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 36, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 17.269683837890625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.36.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 36, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 49.57736587524414, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.36.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 36, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 31.39866065979004, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.36.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 36, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 48.064151763916016, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.36.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 36, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 10.5122652053833, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.36.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 36, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 34.66537094116211, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.36.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 36, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.36.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 36, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 22.800304412841797, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.36.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 36, | |
| "category": "attention", | |
| "edit_norm": 2.099689483642578, | |
| "base_norm": 61.06275177001953, | |
| "rel_norm": 0.03438576583562157, | |
| "changed": true, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.36.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 36, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.36.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 36, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 51.67858123779297, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.36.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 36, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 22.800392150878906, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.37.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 37, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 687.470947265625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.37.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 37, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.88671875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.37.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 37, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 93.49392700195312, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.37.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 37, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 121.9244155883789, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.37.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 37, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 124.55419158935547, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.37.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 37, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 17.879493713378906, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.37.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 37, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 50.63237380981445, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.37.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 37, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 21.809444427490234, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.37.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 37, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 42.81327438354492, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.37.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 37, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 10.926801681518555, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.37.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 37, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 30.795549392700195, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.37.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 37, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.37.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 37, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 21.62625503540039, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.37.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 37, | |
| "category": "attention", | |
| "edit_norm": 2.135906934738159, | |
| "base_norm": 58.566436767578125, | |
| "rel_norm": 0.03646981193707483, | |
| "changed": true, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.37.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 37, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.37.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 37, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 50.35832595825195, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.37.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 37, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 21.626205444335938, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.38.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 38, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 419.9391174316406, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.38.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 38, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.8828125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.38.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 38, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 94.82438659667969, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.38.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 38, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 119.08065795898438, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.38.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 38, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 122.06836700439453, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.38.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 38, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 18.0202579498291, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.38.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 38, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 48.818153381347656, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.38.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 38, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 19.043445587158203, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.38.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 38, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 44.30009460449219, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.38.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 38, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 9.570195198059082, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.38.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 38, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 29.07625961303711, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.38.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 38, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.38.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 38, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 22.145214080810547, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.38.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 38, | |
| "category": "attention", | |
| "edit_norm": 2.3206369876861572, | |
| "base_norm": 57.09755325317383, | |
| "rel_norm": 0.040643370082712994, | |
| "changed": true, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.38.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 38, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.38.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 38, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 50.11199951171875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.38.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 38, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 22.145248413085938, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.39.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 39, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 547.6665649414062, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.39.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 39, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.85546875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.39.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 39, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 93.88056182861328, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.39.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 39, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 126.63441467285156, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.39.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 39, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 130.1889190673828, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.39.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 39, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 17.227842330932617, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.39.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 39, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 48.80883026123047, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.39.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 39, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 20.043716430664062, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.39.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 39, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 51.761051177978516, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.39.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 39, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 13.20450496673584, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.39.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 39, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 26.004425048828125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.39.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 39, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.39.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 39, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 21.77158546447754, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.39.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 39, | |
| "category": "attention", | |
| "edit_norm": 2.494375467300415, | |
| "base_norm": 59.63737487792969, | |
| "rel_norm": 0.04182570866685686, | |
| "changed": true, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.39.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 39, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.39.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 39, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 49.753929138183594, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.39.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 39, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 21.771656036376953, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.4.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 4, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 1513.50830078125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.4.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 4, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.5859375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.4.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 4, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 102.76004028320312, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.4.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 4, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 140.11749267578125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.4.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 4, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 139.9314422607422, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.4.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 4, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 13.606917381286621, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.4.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 4, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.532752990722656, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.4.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 4, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 34.94817352294922, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.4.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 4, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 62.740333557128906, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.4.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 4, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 63.059669494628906, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.4.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 4, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 1959.8887939453125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.4.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 4, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.4.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 4, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 29.85155487060547, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.4.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 4, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 62.46060562133789, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.4.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 4, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.4.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 4, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 59.1715087890625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.4.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 4, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 30.421098709106445, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.40.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 40, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 447.5788879394531, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.40.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 40, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.80859375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.40.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 40, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 94.351806640625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.40.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 40, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 118.92512512207031, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.40.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 40, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 122.05870819091797, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.40.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 40, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.71473503112793, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.40.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 40, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 52.58823013305664, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.40.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 40, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 25.659976959228516, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.40.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 40, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 61.74209976196289, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.40.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 40, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 12.96142864227295, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.40.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 40, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 30.62618637084961, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.40.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 40, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.015625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.40.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 40, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 22.821138381958008, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.40.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 40, | |
| "category": "attention", | |
| "edit_norm": 1.84255850315094, | |
| "base_norm": 57.6536750793457, | |
| "rel_norm": 0.03195908154363315, | |
| "changed": true, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.40.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 40, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.40.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 40, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 52.11252212524414, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.40.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 40, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 22.821218490600586, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.41.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 41, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 411.3920593261719, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.41.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 41, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.4453125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.41.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 41, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 78.41695404052734, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.41.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 41, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 120.59944915771484, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.41.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 41, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 120.42975616455078, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.41.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 41, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 8.183748245239258, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.41.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 41, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 50.77711868286133, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.41.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 41, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 25.880207061767578, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.41.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 41, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 80.70235443115234, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.41.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 41, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 41.015380859375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.41.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 41, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 34.74675369262695, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.41.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 41, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.3810679912567139, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 512 | |
| }, | |
| { | |
| "key": "model.language_model.layers.41.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 41, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 30.81353759765625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2621440 | |
| }, | |
| { | |
| "key": "model.language_model.layers.41.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 41, | |
| "category": "attention", | |
| "edit_norm": 0.10943326354026794, | |
| "base_norm": 62.12864303588867, | |
| "rel_norm": 0.001761397934879307, | |
| "changed": true, | |
| "numel": 10485760 | |
| }, | |
| { | |
| "key": "model.language_model.layers.41.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 41, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.157747268676758, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 512 | |
| }, | |
| { | |
| "key": "model.language_model.layers.41.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 41, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 70.77832794189453, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 10485760 | |
| }, | |
| { | |
| "key": "model.language_model.layers.41.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 41, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 30.813554763793945, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2621440 | |
| }, | |
| { | |
| "key": "model.language_model.layers.5.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 5, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 2444.770751953125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.5.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 5, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.60546875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.5.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 5, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 105.7091064453125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.5.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 5, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 137.9553680419922, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.5.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 5, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 141.8202667236328, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.5.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 5, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 12.598971366882324, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.5.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 5, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 51.98014831542969, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.5.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 5, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.673824310302734, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.5.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 5, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 68.1959457397461, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.5.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 5, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.35024094581604, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.5.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 5, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 6418.32275390625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.5.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 5, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.3810679912567139, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 512 | |
| }, | |
| { | |
| "key": "model.language_model.layers.5.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 5, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 42.208927154541016, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2621440 | |
| }, | |
| { | |
| "key": "model.language_model.layers.5.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 5, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 78.36988830566406, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 10485760 | |
| }, | |
| { | |
| "key": "model.language_model.layers.5.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 5, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 23.157747268676758, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 512 | |
| }, | |
| { | |
| "key": "model.language_model.layers.5.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 5, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 85.52445983886719, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 10485760 | |
| }, | |
| { | |
| "key": "model.language_model.layers.5.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 5, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 43.74039077758789, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2621440 | |
| }, | |
| { | |
| "key": "model.language_model.layers.6.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 6, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 3131.192626953125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.6.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 6, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.55859375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.6.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 6, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 105.70169830322266, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.6.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 6, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 142.10877990722656, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.6.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 6, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 144.1402130126953, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.6.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 6, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 11.691540718078613, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.6.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 6, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 51.019901275634766, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.6.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 6, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 80.14833068847656, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.6.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 6, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 60.51039505004883, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.6.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 6, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 18.86269187927246, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.6.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 6, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 4830.94189453125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.6.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 6, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.953125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.6.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 6, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 30.424535751342773, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.6.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 6, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 66.86538696289062, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.6.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 6, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.6.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 6, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 60.380950927734375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.6.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 6, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 31.235158920288086, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.7.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 7, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 4307.91455078125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.7.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 7, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.671875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.7.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 7, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 104.82913208007812, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.7.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 7, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 142.29483032226562, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.7.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 7, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 142.11972045898438, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.7.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 7, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 13.860440254211426, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.7.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 7, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 54.19287109375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.7.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 7, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 29.18901824951172, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.7.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 7, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 64.74800872802734, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.7.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 7, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 8.974482536315918, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.7.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 7, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 3023.147705078125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.7.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 7, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 2.03125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.7.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 7, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 31.64044761657715, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.7.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 7, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 67.971435546875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.7.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 7, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.6875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.7.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 7, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 62.58329772949219, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.7.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 7, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 31.89817237854004, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.8.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 8, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 3823.255859375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.8.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 8, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.228515625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.8.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 8, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 101.5910415649414, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.8.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 8, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 140.37106323242188, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.8.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 8, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 141.03346252441406, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.8.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 8, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 4.693148612976074, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.8.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 8, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 49.902191162109375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.8.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 8, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 34.3447380065918, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.8.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 8, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 192.17001342773438, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.8.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 8, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 7.74745512008667, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.8.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 8, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 1104.5797119140625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.8.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 8, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.921875, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.8.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 8, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 31.374042510986328, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.8.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 8, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 66.69747924804688, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.8.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 8, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.8.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 8, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 62.25240707397461, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.8.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 8, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 31.720836639404297, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.9.input_layernorm.weight", | |
| "type": "input_layernorm.weight", | |
| "layer": 9, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 1883.884765625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.9.layer_scalar", | |
| "type": "layer_scalar", | |
| "layer": 9, | |
| "category": "scalar", | |
| "edit_norm": 0.0, | |
| "base_norm": 0.4140625, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1 | |
| }, | |
| { | |
| "key": "model.language_model.layers.9.mlp.down_proj.weight", | |
| "type": "mlp.down_proj.weight", | |
| "layer": 9, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 105.72559356689453, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.9.mlp.gate_proj.weight", | |
| "type": "mlp.gate_proj.weight", | |
| "layer": 9, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 136.71664428710938, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.9.mlp.up_proj.weight", | |
| "type": "mlp.up_proj.weight", | |
| "layer": 9, | |
| "category": "mlp", | |
| "edit_norm": 0.0, | |
| "base_norm": 137.72848510742188, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 26214400 | |
| }, | |
| { | |
| "key": "model.language_model.layers.9.per_layer_input_gate.weight", | |
| "type": "per_layer_input_gate.weight", | |
| "layer": 9, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 8.861506462097168, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.9.per_layer_projection.weight", | |
| "type": "per_layer_projection.weight", | |
| "layer": 9, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 53.54083251953125, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 655360 | |
| }, | |
| { | |
| "key": "model.language_model.layers.9.post_attention_layernorm.weight", | |
| "type": "post_attention_layernorm.weight", | |
| "layer": 9, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 26.137964248657227, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.9.post_feedforward_layernorm.weight", | |
| "type": "post_feedforward_layernorm.weight", | |
| "layer": 9, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 104.19583892822266, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.9.post_per_layer_input_norm.weight", | |
| "type": "post_per_layer_input_norm.weight", | |
| "layer": 9, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 15.023260116577148, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.9.pre_feedforward_layernorm.weight", | |
| "type": "pre_feedforward_layernorm.weight", | |
| "layer": 9, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 788.5464477539062, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.layers.9.self_attn.k_norm.weight", | |
| "type": "self_attn.k_norm.weight", | |
| "layer": 9, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 1.9375, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.9.self_attn.k_proj.weight", | |
| "type": "self_attn.k_proj.weight", | |
| "layer": 9, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 30.090221405029297, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.layers.9.self_attn.o_proj.weight", | |
| "type": "self_attn.o_proj.weight", | |
| "layer": 9, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 63.15673065185547, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.9.self_attn.q_norm.weight", | |
| "type": "self_attn.q_norm.weight", | |
| "layer": 9, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 16.5, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| }, | |
| { | |
| "key": "model.language_model.layers.9.self_attn.q_proj.weight", | |
| "type": "self_attn.q_proj.weight", | |
| "layer": 9, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 60.40083694458008, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 5242880 | |
| }, | |
| { | |
| "key": "model.language_model.layers.9.self_attn.v_proj.weight", | |
| "type": "self_attn.v_proj.weight", | |
| "layer": 9, | |
| "category": "attention", | |
| "edit_norm": 0.0, | |
| "base_norm": 30.74000358581543, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 1310720 | |
| }, | |
| { | |
| "key": "model.language_model.norm.weight", | |
| "type": "norm.weight", | |
| "layer": null, | |
| "category": "norm", | |
| "edit_norm": 0.0, | |
| "base_norm": 402.7760925292969, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 2560 | |
| }, | |
| { | |
| "key": "model.language_model.per_layer_model_projection.weight", | |
| "type": "per_layer_model_projection.weight", | |
| "layer": null, | |
| "category": "other", | |
| "edit_norm": 0.0, | |
| "base_norm": 129.72012329101562, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 27525120 | |
| }, | |
| { | |
| "key": "model.language_model.per_layer_projection_norm.weight", | |
| "type": "per_layer_projection_norm.weight", | |
| "layer": null, | |
| "category": "per_layer", | |
| "edit_norm": 0.0, | |
| "base_norm": 14.6741361618042, | |
| "rel_norm": 0.0, | |
| "changed": false, | |
| "numel": 256 | |
| } | |
| ] | |
| } |