Model · llama-7b
LLaMA 7B
Developer: Meta AI
Availability: unknown · checked 2026-09-24 · source
Meta distributed LLaMA 1 weights by application form; no Meta-owned HF repo exists and the form's current state was not checked. Third-party copies (e.g. huggyllama/llama-7b) are re-uploads, not verified against the originals.
Raw record: /data/models/llama-7b.json
Fields
- id
- llama-7b
- developer
- Meta AI
- release_date
- 2023-02-24recorded · source
- weights_status
- open
- availability
- unknown · checked 2026-09-24 · sourcenote: Meta distributed LLaMA 1 weights by application form; no Meta-owned HF repo exists and the form's current state was not checked. Third-party copies (e.g. huggyllama/llama-7b) are re-uploads, not verified against the originals.
- license
- Non-commercial research license; gated access via application to Meta (no standalone public license-text URL located)partial · source
- architecture
- training_data
- Fully disclosed mix: CommonCrawl, C4, GitHub, Wikipedia, Books (Gutenberg + Books3), ArXiv, StackExchange, with exact token-count table per source.recorded · source
- techniques
- transformer-decoderrotary-position-embeddingrmsnormswiglu
- primary_sources
- record_history
- date:2026-09-24 · change:created from primary sources (paper, Meta AI blog post) · by:wilson-pruitt + claude ·date:2026-09-24 · change:availability checked and recorded · by:wilson-pruitt + claude ·
Parents
No edges recorded.
Children
Weights descend
- ← fine_tuned_from Alpaca 7B declared source Stanford CRFM: "fine-tuned from Meta's LLaMA 7B model"
- ← fine_tuned_from Vicuna 7B v1.3 declared source FastChat docs: v1.3 base model 'Llama 1'; model card: 'Finetuned from: LLaMA'. Neither names the size; 7B-to-7B pairing rests on the model name and matching config dims.
- ← adapter_on alpaca-lora-7b declared source Card: 'This repo contains a low-rank adapter for LLaMA-7b fit on the [Stanford Alpaca] dataset.' adapter_config.json: peft_type LORA, r 16, targets q_proj,k_proj,v_proj,o_proj; base_model_name_or_path decapoda-research/llama-7b-hf (a third-party re-upload of LLaMA-7B; the card names LLaMA-7b). The record holds adapter weights only; a runnable model needs LLaMA-7B.
- ← fine_tuned_from Baize v2 7B declared source Card: 'Baize is an open-source chat model fine-tuned with LoRA. This model is a 7B Baize-v2 ... This checkpoint has been merged with LLaMA so it's ready for use.' The LoRA is merged into LLaMA, so the weights descend from LLaMA. Size (7B, so llama-7b) is from the model's name; the README's v1 merge example uses base huggyllama/llama-7b. Intermediate step not recorded: the Baize paper (arXiv 2304.01196, Table 3) has v2 built on Baize v1.5, itself LLaMA-7B fine-tuned, and v1.5 is not recorded in Stemma. This edge is true but skips that step (Wilson's ruling, 2026-09-24: keep, note the gap).
- ← adapter_on guanaco-7b declared source Card: 'open-source finetuned chatbots obtained through 4-bit QLoRA tuning of LLaMA base models'; 'Lightweight checkpoints which only contain adapter weights'; usage loads huggyllama/llama-7b (a third-party re-upload of LLaMA-7B) with adapters timdettmers/guanaco-7b. adapter_config.json: LoRA r 64, targets q,k,v,o,gate,up,down proj; its base path is a local path (/gscratch/zlab/llama/7B). QLoRA trains the adapter against a frozen 4-bit quantized base; the adapter itself is not a quantization of LLaMA, so not quantized_from.
Design
- ← successor_in_series Llama 2 7B declared source Paper: 'Llama 2, an updated version of Llama 1'. Trained from scratch on a new data mix, NOT initialized from Llama 1 weights.
- ← successor_in_series Llama-2-13b-hf declared source Paper: 'Llama 2, an updated version of Llama 1... trained on a new mix of publicly available data', not initialized from Llama 1 weights. Same edge as llama-2-7b.
- ← successor_in_series Llama-2-70b-hf declared source Paper: 'Llama 2, an updated version of Llama 1... trained on a new mix of publicly available data', not initialized from Llama 1 weights. Same edge as llama-2-7b.
- ← design_follows open_llama_7b declared source Card: 'a permissively licensed open source reproduction of Meta AI's LLaMA', trained from scratch on different data (RedPajama, not LLaMA's original mix). No weights passed.